Recently, two very different but equally thought-provoking articles were published within a day of each other on the Bioethics Today blog. After reading both, I found they were like two sides of the same coin when thinking about disability ethics. Wolfe and Escandon wrote about a difficult surrogacy story involving hypoplastic left heart syndrome, or HLHS, and described hypoplastic left heart syndrome (HLHS) mainly through suffering, burden, and low quality of life. Elster, Parsi, and Caplan called out political rhetoric that portrayed disabled lives through dependency and low possibility. Of course, the contexts and ethical questions are not the same. But reading the articles, one after the other, left me wondering about the messages bioethics sometimes sends about disability.

I wondered what a person living with HLHS might think about when reading an account of a life like theirs, framed by burden and premature death. In their article, Wolfe and Escandon’s descriptions of HLHS, its lifesaving interventions, and a life beyond those interventions were understandably bleak. While such a portrayal is reasonable given the medical challenges of HLHS, it may inadvertently suggest to the disability community that lives shaped by disability or complex medical needs are inherently less valuable or fulfilling.

It made me wonder: If ableist portrayals of disability deserve scrutiny when they shape public attitudes and policy, should they receive any less scrutiny when similar medical tragedy models are communicated by bioethicists whose expertise may influence how families imagine the possibilities of a disabled child’s life?

A recent case in Texas involved an Alaska surrogate and intended parents in California. Reports stated that their agreement contained a termination clause addressing fetal abnormalities. At about twenty weeks, the fetus was diagnosed with HLHS.  The Associated Press, reporting on court documents, revealed the intended parents sought termination of the pregnancy after medical consultations, but the surrogate later refused.

HLHS is a serious diagnosis, and it requires staged surgeries and lifelong cardiac care. But advances in surgical and specialized cardiac care have substantially improved outcomes, with one-year survival increasing from 43% in the 1980s to 80% in 2023. In fact, a recent study from the Children’s Hospital of Philadelphia (CHOP) that followed patients who underwent staged surgeries from 1984 to 2023 found that most surveyed adults reported good-to-excellent general health and quality of life. Ultimately, the diagnosis of HLHS is serious, but diagnosis alone cannot establish that the resulting life will have an unacceptable quality of life.

In their article, Wolfe and Escandon understandably emphasized the seriousness of HLHS and the burdens of treatment. They wrote that “surviving patients have limitations in quality of life,” and that some loving parents may conclude that the treatments and risks are “out of proportion to any of the potential benefits.” Proportionality is a valid clinical ethics concept that weighs an intervention’s benefits against its burdens. But when treatment proportionality migrates into what I would call “existential proportionality,” it creates a bioethical tension. (I use that phrase descriptively, not as an established clinical doctrine). It goes from “Are the burdens of the surgery proportionate to the benefits?” to “Are the possible burdens of this child’s life proportionate to the benefits of that life?” The second question projects an unknown future and assigns a moral weight to disability before life has begun.

What would a person living with HLHS hear in an ethical analysis questioning if their life has a “proportionate benefit” to its burdens? They may ask whether the language of proportionality can overlook the values we bring to judgments about disability.

Wolfe and Escandon also recognized that, after birth, parents may choose among medically and ethically permissible options, including staged palliation, transplantation, and comfort care. Comfort care can be an ethically appropriate choice for a newborn with HLHS because the condition may be accompanied by significant genetic, neurologic, or other extracardiac conditions that can alter prognosis. These comorbidities may make the burdens of staged palliation disproportionate to its potential benefits. But that is different from projecting the burdens or value of a child’s future life from an HLHS diagnosis itself.

The subsequent course of this Texas case makes it relevant. Born on August 12, the baby was determined to be eligible for surgery. He underwent the Norwood procedure on August 17, and he was reported to be recovering–but in serious condition. None of this treatment guarantees his future or lessens the serious issues associated with his diagnosis. But this case does invite an important question: if uncertainty about future suffering and treatment burdens weighs ethically in favor of comfort care, shouldn’t the other side of that uncertainty (like the possibility of survival and a life) receive comparable weight?

Disability scholars have cautioned for many years that reproductive and medical decisions do not happen in a value-neutral information environment. Adrienne Asch described the tendency to allow a disability diagnosis to represent a whole person. She and David Wasserman used the rhetorical device called Synecdoche, where one characteristic becomes predictive of an entire future. Joseph Stramondo has also argued that tensions between disability bioethics and mainstream bioethics are, in part, differences in the assumptions non-disabled people make about the lived experiences of disabled people. The concerns are again reinforced when we hear about the “disability paradox” where people living with disability often report a better quality of life than their nondisabled observers.

Altogether, this scholarship gives reason to scrutinize descriptions of a future life as burdensome, dependent, suffering, etc. Empirical studies of medicine can estimate morbidity and mortality, but they cannot determine the burden of a disabled life. That is a value judgement based on one’s own assumptions about a future person whose life, relationships, adaptations, and happiness remain unknown.

The issues described in this essay are why the contrast of the two articles in the Bioethics Today Blog matters. If assumptions matching disability with diminished human value are receiving scrutiny when they appear in public health policy, then they should not be invisible when they come up in reproductive or neonatal decision-making as well.

Kyoko L. Yoda, DMD, MSPH, MSD is an instructor at Children’s Hospital Colorado.

The post When Bioethical Neutrality is not Neutral: Disability, Autonomy, and the Value of Life appeared first on Bioethics Today.

In early August, an Alaska woman named McKenna West was serving as a gestational surrogate for a Los Angeles couple when a prenatal diagnosis revealed the fetus had hypoplastic left heart syndrome, a serious but treatable congenital heart defect. According to West’s own account, the intended parents asked her to terminate the pregnancy. She refused and traveled to Texas, where abortion is illegal, to give birth.

Texas Attorney General Ken Paxton then intervened directly, securing a court order compelling stabilizing and life-sustaining care for the newborn, whom West calls Baby Gabriel, over the intended parents’ objection. The baby underwent surgery in August, and Paxton’s office withdrew from the case once treatment was secured.

Texas law does not consistently recognize fetal personhood. It restricts abortion almost entirely yet stops short of granting a fetus independent legal standing in other contexts, such as inheritance or tort claims. That ambiguity is not an oversight. It is the compromise that lets the state regulate pregnancy without fully committing to the legal consequences of calling a fetus a person.

Paxton’s intervention and interpretation came on exactly where it was needed to win, and nowhere else. To override two parents’ medical decisions and compel a specific surgical outcome, the state needed the child to already function as a rights-bearing individual with interests independent of his parents’ wishes.

This matters because the same legal architecture offered no comparable protection to the person who actually made this outcome possible. West carried this pregnancy, refused a request to terminate it, and left Alaska to secure care for a child who was not genetically hers.

Under the American Society for Reproductive Medicine’s 2023 ethics committee opinion, gestational carriers are the sole source of consent for their own medical care from embryo transfer through delivery and aftercare, specifically because complications can create situations where fetal or neonatal wellbeing might otherwise be weighed against the carrier’s own health. Her autonomy, in other words, is the one interest this case should have made legally uncomplicated.

This is nothing but selective personhood: not hypocrisy in the simple sense, since nobody in this case lied about what they believed, but a legal reflex where fetal personhood gets invoked exactly when it produces the desired outcome and recedes when invoking it consistently would mean granting pregnant women broader protection than the state is willing to extend. Bioethics has spent decades building frameworks for maternal-fetal conflict, weighing a pregnant woman’s autonomy against a fetus’s interests when the two disagree. None of that literature anticipated a third actor, a state official, asserting fetal personhood only when the pregnant woman’s wishes and the fetus’s interests happened to align, and staying silent every other time.

That alignment itself is worth pausing on, rather than treating it as simply fortunate. It is not a coincidence that this became the case Texas chose to fight for. CNN reporting found that West was connected to Alliance Defending Freedom, a conservative Christian legal organization now representing her, through Live Action, an anti-abortion nonprofit that paid for her travel from Alaska to Texas.

 ABC13 Houston reported that the case has drawn in Texas Alliance for Life and Texas Right to Life as well, since it touches three priorities those groups already share: opposing abortion, defending parental rights, and restricting commercial surrogacy. A law professor quoted by CNN noted the case also dovetails with a broader anti-surrogacy push among conservatives. None of this required West to be doing anything but telling her own story, but it explains why the state’s selective personhood claim found a case where its logic and a sympathetic set of facts happened to line up perfectly, and why that alignment should not be mistaken for evidence that the underlying legal reasoning is any more consistent than it was before.

If bioethics actually wants and aims to close this loophole rather than just name it, the fix has to be structural. A state that invokes fetal personhood to justify overriding one decision should be bound to apply that same personhood status consistently wherever else it carries legal weight, in inheritance claims, in tort law, in every context where personhood already does real work, or explain specifically why it doesn’t. A consistency requirement like this would not settle the underlying disagreement over what constitutes a fetus. It would simply stop states from having it both ways, restricting a pregnant woman’s choices on the premise of personhood while denying that same fetus personhood everywhere else it would be inconvenient to grant it.

Until that requirement exists, every state that intervenes this selectively will keep calling it protection. The next carrier who finds herself standing where McKenna West stood deserves a bioethics literature that saw this coming, not one still catching up to it after the surgery is already over.

Yodhin Aggarwal, MHA

The post Personhood, When Convenient? appeared first on Bioethics Today.

Lindsay Clancy is a former labor and delivery nurse who is on trial for the January 24, 2023 strangulation deaths of her three children. Clancy did not have a documented significant psychiatric history, but over a four-month period at the end of 2022, she developed significant depression, anxiety, insomnia, and suicidal ideation. She sought help multiple times through psychiatrists, mental health providers, hospitalizations, calls to suicide hotlines, and was prescribed over a dozen psychiatric medications during that period. Her condition worsened and culminated in the deaths of her children and a suicide attempt that left her paralyzed. Her defense argues that she was suffering from undiagnosed bipolar disorder, postpartum depression, and postpartum psychosis when she killed her children.

Postpartum psychosis is a rare but severe psychiatric condition associated with childbirth. It can involve mania, severe depression, delusions, hallucinations, confusion, agitation, and rapidly changing mental states. Recent expert consensus estimates that it occurs after roughly 0.1–0.2% of deliveries among women without a prior psychiatric history. Postpartum depression affects many more women: 1 in 7.

During the trial, Clancy’s defense attorney Kevin Reddington asked a psychiatrist a deceptively simple question:

Is postpartum psychosis in the DSM?

The exchanges were striking. Dr. Alia Goodheart, a psychiatrist who had treated Clancy at McLean Hospital, testified that she did not know exactly where postpartum psychosis appeared in the DSM-5. Asked whether she had ever seen it there, she said she did not recall paying attention to where it was categorized.

Later in the trial, psychiatrist Dr. Donald Condie was questioned about the same manual. Asked what a reader would find by looking for postpartum psychosis, he testified that there was “one mention on page 148.”

Then, on Day 18 of the trial, psychiatrist Dr. Avram Mack (testifying for the prosecution) explained that postpartum psychosis is not a standalone disorder found in the DSM-5. Instead, clinicians generally diagnose an underlying psychiatric disorder and, where appropriate, use a “with peripartum onset” specifier.

Attorney Reddington goes straight at this omission: “Are you aware, sir, that postpartum psychosis and postpartum depression are pretty much virtually not even mentioned in your DSM-5-TR?…It’s a paragraph. Out of a thousand pages, it’s a paragraph.”

Whatever conclusions one reaches about Clancy’s mental state or criminal responsibility, these exchanges expose a question that extends far beyond one defendant and one courtroom:

Does our psychiatric vocabulary adequately describe what can happen to women after childbirth?

The answer is less straightforward than patients—or perhaps even many clinicians—might reasonably expect. Postpartum depression and postpartum psychosis are disorders that are commonly talked about, but the DSM does not independently diagnose. It is true that they can be classified as a sub-species of major depressive disorder or psychosis (via a specifier of peripartum onset), but they do not exist as standalone diagnoses. That distinction may sound technical. But diagnostic categories shape what clinicians are trained to recognize, what researchers study, what patients are told to watch for, and what symptoms are likely to be understood as part of a coherent syndrome. Classification affects visibility. And visibility matters. Clancy sought help from at least six different mental health providers, visited an emergency room, requested care at a women’s and infants’ hospital, checked herself into an inpatient stay voluntarily, called suicide hotlines twice, and received over 32 prescriptions—all in the span of 4 months. Not one of these encounters offered Clancy an accurate diagnosis or relief from her reported symptoms.

The four-week problem

There’s another problem that may have played a role in Clancy not receiving a diagnosis of postpartum psychosis and/or depression: DSM’s peripartum-onset specifier has traditionally applied when onset occurs during pregnancy or within four weeks after childbirth.

Four weeks.

That means a qualifying mood episode beginning 27 days after delivery may formally receive a peripartum specifier, while a comparable episode beginning later may not. Clancy’s last delivery had been roughly 5 months prior (baby Callen was 8 months old at the time of his death and her older children were 3 and 5 years old), not four weeks.

Reddington read the DSM’s peripartum-onset specifier out load to Dr. Mack during his testimony: it applies to mood episodes beginning during pregnancy or within four weeks after delivery. Dr. Mack confirmed, “That’s the four-week cutoff that’s on that specifier.” Reddington immediately responded sarcastically, “When was this last amended? 1958?”

The DSM’s approach becomes especially striking when compared with contemporary obstetric guidance. The American College of Obstetricians and Gynecologists addresses mental health conditions that may first arise during pregnancy or during the first year postpartum, and its clinical guidance explicitly includes postpartum psychosis.

Diagnostic categories are choices, not discoveries: other systems made a different choice

The DSM (published by the American Psychiatric Association) is not the only way to organize psychiatric illness.

The World Health Organization’s ICD-11 (used more outside of the U.S.) handles things differently. It has an explicit category called “Mental or behavioural disorders associated with pregnancy, childbirth or the puerperium, with psychotic symptoms” (6E21), alongside a corresponding category without psychotic symptoms (6E20). WHO’s 2024 CDDR is intended precisely as clinical diagnostic guidance, not merely a statistical coding manual.

A major comparison of ICD-11 and DSM-5 explains these differences. The ICD framework reflects, in part, the language used by obstetricians and other clinicians who commonly describe conditions as postpartum depression or postpartum psychosis, while the DSM tradition has emphasized the underlying psychiatric syndrome.

But their disagreement reveals something important: diagnostic manuals do not merely discover categories that exist in nature. Human beings decide how illness should be divided, named, and recognized.

Those choices have consequences.

A patient herself does not experience disciplinary or geographic boundaries. She experiences symptoms. A healthcare system should make those symptoms easier to understand, not harder.

Psychiatry itself is reconsidering the question

This concern is not an argument invented by attorneys in a criminal trial. Researchers are actively debating it.

A recent international expert consensus statement argues that postpartum psychosis has a sufficiently distinctive timing, presentation, prognosis, treatment response, and relationship to bipolar-spectrum illness to justify clearer classification. The authors recommend recognizing postpartum psychosis as a distinct diagnostic category and propose an onset window of up to 12 weeks after childbirth.

That proposal is not yet the DSM standard. But its existence should make us ask whether our current categories are doing enough. As I write this, the next version of the DSM (DSM-6) is in development, which presents an opportunity for revisions.

The Clancy case should not define postpartum psychosis

There is an important ethical danger here. Postpartum psychosis should not become publicly synonymous with infanticide or filicide. The Clancy case presents extraordinary and contested facts. Diagnosis and criminal responsibility are different questions, and experts may disagree about both.

If the only time Americans hear the term “postpartum psychosis” is after a horrific tragedy, increased awareness could easily become increased stigma. Women should not need a catastrophe before medicine examines its categories.

We need better research, clearer communication between obstetrics and psychiatry, more education for families and frontline clinicians, and diagnostic systems willing to evolve when their categories no longer serve patients well.

Jennifer Blumenthal-Barby, PhD, MA

AI statement: I used ChatGPT to go back and forth on blog post content, but all ideas, arguments, and writing of final draft in this blog post are mine.

All views expressed in this essay are entirely those of the authors and not those of any of the author’s affiliations

The post In the News… When Psychiatry Fails Women: Lessons from the Lindsay Clancy Trial appeared first on Bioethics Today.

As individuals concerned with ethics in science, medicine, and public health, we wish to express our dismay at the approach taken by Senate leaders to pursue an unwarranted, ill-advised, and frankly bizarre vendetta against a scientist, Dr. Anthony Fauci, rather than making a sincere effort to learn from the painful experiences of the pandemic to improve national preparedness. No human is perfect, but Dr. Fauci deserves appreciation for his decades of service to American science, and he deserves respect for taking on the difficult work of advising federal policy makers during the worst plague of this century. We condemn the decision to hold him legally in contempt of Congress as the outcome of a staged public persecution, and we fear the future implications of politically and legally targeting scientists who advise the government.

Policy makers have an ethical obligation to gather information and learn from experience, and there is much to be learned from the pandemic response, both good and bad. There are legitimate ethical and scientific questions to be asked about the safety and utility of viral gain of function research, the effectiveness of various public health policies, and more. But bringing Dr. Fauci to the Senate under conditions of repeatedly threatened prosecution, thereby forcing him to invoke his fifth amendment rights, serves neither to obtain information nor learn lessons.

If the Senate wished to hear from Dr. Fauci and learn how to be better prepared for the next pandemic, they could have provided him with Use Immunity, and he would not have invoked his 5^th Amendment right to decline to answer questions, even in the face of prior threats of prosecution. But they chose not to do so, showing that neither protecting the public from future pandemics nor learning were the aims of the hearing.

Some Senators instead used the hearing as a blunt warning to other physician-scientists to avoid public service or risk being ensnared in a political and legal trap like the one laid for Dr. Fauci.

Dr. Fauci’s lifetime of service is well-documented. He served 7 U.S. presidents—Republicans and Democrats alike. He led the National Institute of Allergy and Infectious Diseases (NIAID) for nearly 4 decades. He has received numerous honors, including honors from President Trump for his efforts on the COVID vaccine program, Operation Warp Speed. His career has been built on research integrity, pursuing answers to important questions, and an unwavering commitment to bettering all people’s lives.

During the COVID-19 pandemic, Dr. Fauci was, as always, an advisor to policymakers, but he also became the public face of the federal government’s response. As such, he fought to maintain credibility for sound science and evidence-based COVID responses, sometimes eliciting the ire of those with differing views. But as he had done throughout his long career, he worked with a wide range of scientists and sought to make the best recommendations he could to policymakers under evolving, uncertain, and frightening conditions. 

The proper way to investigate the origins of COVID, or any other question about the pandemic and the nation’s responses to it, is through rigorous scientific inquiry, which can then inform policies. Staged, political hearings with predefined scapegoats teach nothing, and using criminal law rather than scientific inquiry to answer scientific questions is damaging.

Signatures,

List of 130 signatories can be found here: Fauci Support Letter (Responses)

(University titles and other affiliations are for identification only, no institutional endorsement is made or implied by signatories)

Arthur Caplan¹, Matthew Wynia², Dominic Sisti³, Lainie Ross⁴

1. New York University, Grossman School of Medicine
2. University of Colorado, Center for Bioethics and Humanities
3. University of Pennsylvania, Perelman School of Medicine
4. University of Rochester Medicine
   

Corresponding author: Matthew Wynia, University of Colorado Center for Bioethics and Humanities

(Photo Credit: NIH Image Gallery from Bethesda, Maryland, USA, Public domain, via Wikimedia Commons)

The post An Open Letter of Support for Anthony Fauci from the Bioethics Community appeared first on Bioethics Today.

What if, instead of another conference panel, bioethicists hosted an exhibition, curated a playlist, staged a performance, or collaborated on a short film?

What if our work did more than explain and evaluate ethical problems? What if it also moved people, unsettled assumptions, and made unfamiliar experiences more difficult to ignore?

Our field has developed powerful tools for identifying values, clarifying concepts, evaluating arguments, and guiding decisions. It is also an applied field, with bioethicists regularly working alongside clinicians, researchers, policymakers, patients, and communities.

These analytic and practical contributions remain essential. But they are not enough.

Bioethics should become more open to artistic and creative practices, not simply as ways of communicating conclusions already reached, but as methods for exploring ethical questions.

Old Habits in a Young Field

Modern bioethics emerged largely within universities, hospitals, and research institutions. Unsurprisingly, it adopted their dominant forms of knowledge production: scholarly articles, books, conference presentations, lectures, reports, policy briefs, and committee recommendations.

These forms work. They allow us to develop careful arguments, scrutinize evidence, and communicate with professionals and decision-makers. Bioethicists have become particularly adept at meeting scientists, clinicians, and policymakers on their own institutional terrain, providing the arguments and documents needed to reshape policy or practice.

Yet these forms also impose limits. They tend to privilege explicit argument over sensory experience and abstraction over emotion. Their audience is primarily professional. Even when bioethicists study decisions that profoundly affect patients, research participants, caregivers, and communities, our work often remains inaccessible to many of them.

Making an academic argument shorter or adding an infographic does not necessarily change how the ethical issue itself is understood. Creative practices can.

Art as a Method of Bioethics

Art is often treated in academic settings as illustration, enrichment, or outreach, something added after the serious intellectual work has been completed. That misses its ethical and epistemic potential.

A story can place an audience inside a conflict rather than leave it observing from a distance. A photograph can make structural exclusion visible. Theatre can expose the competing responsibilities embodied in a clinical encounter. Music, dance, and poetry can communicate vulnerability, dependence, fear, grief, or moral distress in ways that resist reduction to propositions.

Such works do not merely generate feelings about ethical issues. They can reveal dimensions that conventional analysis has overlooked. They can shift whose experiences become visible, what counts as relevant knowledge, and which questions appear worth asking. They can also support the agency and presence of people too often relegated to the role of passive subjects.

This is not an argument against analysis. Creative work without critical reflection can reproduce stereotypes, sentimentalize suffering, or substitute emotional impact for ethical judgment. But conventional bioethics also has characteristic failures. It can flatten lived experience, treat social power as contextual background, and assume that what can be clearly stated is all that matters ethically.

The medical and health humanities have long treated literature, visual art, performance, and other creative practices as credible and legitimate ways to understand illness, care, disability, and professional life. These fields developed in part because analytic arguments alone could not capture the full complexity of health and illness, and bioethics has much to learn from both their successes and their challenges.

The point is not to choose between creative and analytic approaches, but to recognize that each can correct some of the other’s limitations.

Creating Institutional Space

For more than a decade, the Canadian Journal of Bioethics / Revue canadienne de bioéthique has invited visual art, fiction, poetry, film criticism, and other creative works through its Arts, Culture and Creative Works section. Other journals have created related spaces for non-traditional forms of bioethical inquiry. Narrative Inquiry in Bioethics, for example, explores bioethical questions through personal stories, qualitative research, and case studies that leverage rich descriptions of human experience to help deepen ethical understanding.

In 2016, an Art + Bioethics project paired emerging artists with junior bioethics scholars, culminating in both a gallery exhibition and a journal collection. The point was not to decorate articles with attractive images. Artists and bioethicists worked together on shared questions, bringing different methods, assumptions, and forms of attention to the collaboration. The artworks and bioethics texts informed one another, shaping both the final works and their meaning.

Similar possibilities are emerging in bioethics education and graduate research. Students increasingly use graphic narratives, speculative fiction, visual media, performance, and participatory artistic activities to investigate ethical questions and engage communities. Their work shows that creative practices can be integrated into serious conceptual and empirical research rather than relegated to extracurricular communication.

But isolated experiments are not enough. Journals, graduate programs, research centers, and funders must create legitimate space for this work.

Creativity Still Requires Rigour

Opening bioethics to artistic practice requires standards appropriate to different forms of inquiry.

This challenge became clear in a 2016 research project entitled Responsible Conduct in Research-Creation: Providing Creative Tools to Meet the Challenges of an Emerging Field. Many conventional research norms had emerged from the empirical sciences and did not translate neatly into artistic settings, including forms of “punk ethics” intended to destabilize audiences. At the same time, artistic freedom did not eliminate responsibilities concerning consent, collaboration, attribution, power, harm, or accountability. The resulting research-creation toolbox treated differences in methods and objectives as matters requiring explicit discussion and ethical judgment, rather than the imposition of a general set of rules.

Creative bioethics requires the same seriousness. Who has the authority to represent an experience? Who controls the resulting work? How are collaborators credited? When does provocation become exploitation? How should audiences participate? What obligations arise when a project makes suffering visible?

Art can empower, but it also creates new forms of ethical responsibility for bioethicists.

Expanding What Bioethics Can Do

Not every bioethicist should become a painter, novelist, or performer. That would be both unrealistic and exhausting. The more reasonable invitation is to collaborate and experiment.

Journal editors can create space for creative scholarship and recruit reviewers able to evaluate it fairly. Supervisors can support students who want to integrate artistic practices into their research. Funders and universities can recognize exhibitions, performances, films, graphic works, and community-based creations as substantive research contributions. Practising bioethicists can consider whether stories, images, performance, or participatory activities might enrich engagement with patients, professionals, and communities.

The people affected by bioethical decisions do not primarily encounter them through journal articles. They understand ethical problems through bodies, relationships, images, stories, institutions, and emotions, as well as through arguments.

Bioethics already asks how we should live together amid vulnerability, technological change, unequal power, and competing ideas of the good. Creative practices allow us to encounter those questions differently. They can make abstract issues tangible, disrupt familiar categories, and bring neglected experiences into view.

Bioethics should continue to analyze, deliberate, and advise. But it should also make, perform, imagine, and create.

We should not only explain why ethics matters. We should help people see, hear, and feel what is at stake.

Bryn Williams-Jones, PhD, is professor and director of the Department of Social and Preventive Medicine, School of Public Health, Université de Montréal.

The post Making Bioethics Creative: Art Should Be Treated as Method, Not Decoration appeared first on Bioethics Today.

This editorial appears in the August Issue of the American Journal of Bioethics

“Better babies.” “Fitter families.” “Survival of the fittest.” “Three generations of imbeciles are enough.” These phrases are not merely historical reminders of the United States’ regrettable eugenic past but are appearing in an increasingly eugenic present. Eugenics may have seemed dormant, but has recently been reawakened by the alt-right, tech billionaires, and figures such as Robert F. Kennedy (RFK), Jr., Stephen Miller, and President Trump. Autism has become the most recent target of eugenic ideology.

The administration’s stoking fear of and offering feigned support for autistic children and their families has bolstered MAHA’s ongoing eugenic rhetoric about the “scourge” of autism. This harmful language continues despite scientific evidence supporting MAHA’s misplaced (and scientifically refuted) views about the causal roles of vaccines and Tylenol. Disastrous political ideologies of the 19th century dominate this administration’s policies, but none so perniciously as the ideology of eugenics as applied to autism.

Ridding society of disabled individuals has long been the goal of eugenics. In fact, “[t]he goal of the Nazi Euthanasia Program was to kill people with mental and physical disabilities” (Holocaust Encyclopedia). This grim eugenic history dates back to the 19th century when it was considered part of mainstream medicine and science. Scholars have argued that medicine legitimized eugenics through “expertise laundering,” whereby “unsettled expertise [is laundered] through various interdisciplinary ‘transactions’ until it appears settled”. From eugenics to Lysenkoism to MAHA, ideologies are legitimized by those in power and authority asserting a goal of improving society yet, in reality, harming public health, demonizing vulnerable minorities, and ignoring individual dignity and well-being.

Public health and eugenics share similar nomenclatures. Both focus on society, prevention, and betterment. The difference between the two, however, is in intent. Public health seeks to improve the lives of individuals by improving the lives of the community. This contrasts with the crude consequentialism of eugenics, which seeks to eliminate the lives of some to decrease economic burdens, thereby purportedly improving the lives of many. Eugenics seek to serve the powerful; public health seeks to empower the marginalized.

Historically, those promoting eugenics have often couched their agenda in terms of public health, which is precisely what RFK, Jr. and his MAHA cronies have been doing through their rhetoric around autism. They are utilizing a public health rationale to peel back decades of laws and regulations aimed at increasing vaccination rates to reduce communicable disease globally with one narrow and bogus goal: eliminating autism. This is problematic because reducing vaccination rates will not eliminate autism. It will, however, greatly harm many individuals while endangering public health. And it will ultimately lead to an increase rather than decrease in morbidity and mortality—exactly what MAHA claims it wants to prevent.

The U.S. eugenics movement of the early 20th century served as the blueprint for the Nazi regime that espoused a similar mission of exterminating the disabled. In the oft-quoted Supreme Court case of Buck v. Bell, which has notably never been overturned, Justice Oliver Wendell Holmes, Jr. asserted that

It would be strange if it could not call upon those who already sap the strength of the State for these lesser sacrifices, often not felt to be such by those concerned, in order to prevent our being swamped with incompetence. It is better for all the world if, instead of waiting to execute degenerate offspring for crime or to let them starve for their imbecility, society can prevent those who are manifestly unfit from continuing their kind.

This claim by Holmes continues to have life among contemporary leaders, highlighting the fact that eugenics in America has never really gone away, but merely has been hiding in plain sight. For instance, RFK, Jr. erroneously describes autism as a monolith, describing it in its most severe expression, which notably accounts for about a quarter of autistic individuals. He makes his assertions about autism using the language of public health by referring to autism as an epidemic, based on more faulty reasoning about the increasing number of autism diagnoses. The tenor and tone of his rallying cry eerily echo the explicitly eugenic language of Holmes:

Autism destroys families. More importantly, it destroys our greatest resource, which is our children. These are children who should not be suffering like this … These are kids who will never pay taxes, they’ll never hold a job, they’ll never play baseball, they’ll never write a poem, they’ll never go out on a date. Many of them will never use a toilet unassisted. And we have to recognize we are doing this to our children. and we need to put an end to it.

Both statements rely on the language of public health and public welfare to mask economic and ableist motivation. In describing the ableist notions that fed into the Nazi elimination of 300,000 disabled individuals, historian Dagmar Herzog writes that both mental impairment and/or mental illness

destabilize the idealized notion of the autonomous subject and, not least and most enduringly, because of the challenge that these impairments have raised for the expectation of being able to contribute labor to one’s community, rather than requiring the labor of care and support from others.

Interestingly, both Holmes and Kennedy discuss vaccination—Holmes uses it as a metaphor, whereas Kennedy blames vaccinations for causing autism as part of his public health subterfuge. Both assert the state’s power and authority to promote the public welfare; however, neither acknowledges the significant difference between communicable disease and disability, with one being contagious and capable of rapid spread to many and the other being a much more amorphous concept that can be inborn, acquired, or simply created as a medical condition. The interest in the state in stemming communicable diseases such as measles or COVID-19 is in the prevention or eliminating a disease vector rather than the prevention of the existence of a human being.

A 1997 American Journal of Public Health article comparing and contrasting eugenics and public health in the wake of the Human Genome Project reasons that “[s]ome of what was done in the name of eugenics was also done in the name of infection control and public health. Eugenics was not an isolated movement whose significance is confined to the histories of genetics and pseudoscience”. The conclusion is relevant to the current debate about autism “prevention”: “Past similarities between eugenics and public health serve as an alarm clock for all the health sciences”.

The response to disability should never be the elimination of people with disability but should rather be a collective response to helping those individuals participate as fully in our communities as possible. Autism is not the scourge on society that RFK, Jr. would have us believe; rather, it is rampant elite ableism that is the true epidemic. In fact, one could argue that an ableist classism has animated much of the policy decisions to eliminate vaccine recommendations—tools that in the absence of a universal health care system have greatly benefited marginalized groups. The tools and language of public health would be better applied in preventing and reducing the burden of this narrowmindedness rather than being bastardized to promote harmful, dangerous, and bigoted ideology.

Nanette Elster, JD, MPH, Kayhan Parsi, JD, PhD, HEC-C, and Art Caplan, PhD

The post Laundering Public Health: Using Autism to Revive Eugenics appeared first on Bioethics Today.

This essay is a part of our “In the News…” series, where bioethicists give their take on popular news stories.

 The medical record rarely makes headlines. In the Lindsay Clancy murder trial, however, the medical record has become part of the evidence used to understand the mental state of a mother accused of killing her three children.

In January 2023, Clancy was accused of strangling her three children in the family’s Massachusetts home before attempting suicide. Clancy survived the suicide attempt and has pleaded not guilty to murder by reason of insanity. Her trial began on July 27, 2026, in which her defense argued that she was experiencing severe mental illness, including postpartum psychosis, and therefore was not criminally responsible for her actions. The prosecution has argued that the killings were deliberate and premeditated. Consequently, testimony about Clancy’s psychiatric symptoms, diagnoses, medications, and treatment in the months preceding the deaths has become central to the trial, with numerous medical professionals testifying.

Among the many clinicians whose treatment of Clancy has been examined during this trial is psychiatrist Jennifer Tufts, who had treated Clancy for months leading up to the alleged murder. On August 10, Tufts was cross-examined about her treatment of Clancy in the months preceding the deaths of her three children. During questioning, Clancy’s defense attorney, Kevin Reddington, referred to a note in which Tufts documented “pressured speech.” Tufts testified that Clancy did not have pressured speech; however, her note stated “pressured speech.” When Reddington challenged Tufts about the discrepancy between her testimony and the language in her note, Tufts responded, “I don’t care what it says. I know what I meant.” She subsequently clarified that she had written the note but maintained that Reddington had misinterpreted it.

The exchange illustrates an ethical issue that extends well beyond this particular case and serves as a great reminder of clinicians’ responsibility to document accurately and clearly in the medical record.

Clinical documentation is often discussed in terms of accuracy, completeness, billing, liability, or regulatory compliance. These are important considerations, but they do not exhaust the ethical significance of the medical record. Documentation is also a form of clinical communication. Information gathered and interpreted by one clinician becomes available to others who may subsequently participate in a patient’s care. The record, therefore, carries an ethical obligation similar to other forms of professional communication: information should be represented truthfully, with sufficient precision for its clinical purpose, and in a manner that does not create a misleading impression.

This creates an important distinction between what a clinician intended to communicate and what the record actually communicates. A clinician may know what was intended when a note was written. A subsequent clinician does not have access to that intention. The subsequent clinician has access to the words documented in the record. If those words reasonably communicate something different from what the author intended, the discrepancy becomes clinically and ethically significant.

The ethical obligations associated with clinical documentation can be understood through several familiar principles. Veracity requires clinicians to represent clinical information truthfully. Nonmaleficence requires consideration of the potential harms that may result when inaccurate or misleading information is introduced into a patient’s record and subsequently relied upon. Fidelity is also relevant because patients and clinicians depend upon the medical record to communicate information across professional and temporal boundaries. Documentation, therefore, participates in relationships of trust not only between clinicians and patients but also among members of the healthcare team.

The Clancy trial provides an unusually visible illustration because the medical record is now being examined in a courtroom. Yet the ethical issue does not depend upon litigation. Every clinical record is potentially a handoff to someone who was not present for the original encounter. The author may not be available to explain what a particular phrase meant, and the patient should not bear the consequences of ambiguity that could reasonably have been avoided through more precise documentation and attention to detail.

The relevant question for clinicians, therefore, is not simply whether they know what they meant when documenting an encounter. The more important question is whether another reasonable clinician reading the record would understand what was meant. If the answer is uncertain, the responsibility rests with the author to improve the documentation rather than with the subsequent reader to reconstruct the author’s intention.

Documentation is part of patient care. By entering information into the medical record, clinicians create a representation of a patient’s clinical history that will inform the work of others. That representation carries ethical obligations of accuracy, clarity, and fidelity to what was actually observed and assessed. “I know what I meant” may explain a clinician’s intention, but intention alone does not determine what a medical record communicates. The ethical responsibility of clinical documentation is to ensure that the record communicates the patient’s clinical reality as accurately and clearly as possible without room for interpretation, including to those who will read it after the author is no longer present to explain it.

The Clancy trial highlights the potential consequences of imprecise clinical documentation and offers a broader wake-up call for clinicians: a medical record may one day be scrutinized far beyond the clinical setting. Clinicians cannot know whether a note will be read by another member of the care team years later, examined in litigation, or, as in this case, displayed before the public and scrutinized in a courtroom. The ethical obligation is therefore not simply to know what was meant, but to document it accurately and clearly enough that the record does not require the author to explain what the words were intended to mean.

Mariah Chobany, PhD, HEC-C is a Clinical Ethicist at Sanford Health

The post In the News…“I Know What I Meant:” The Ethical Responsibility of Clinical Documentation appeared first on Bioethics Today.

A gestational surrogate “fleeing” from Alaska to Texas is getting a lot of press. The issues around this story may not be that new, but rather reflect the precarious situation between legal surrogate contracts and the primary rights of the surrogate as a patient. This case raises several ethical challenges in commercial surrogacy, as well as emerging clinical ethics issues in fetal health centers. Frequently, the surrogate pregnancy is the result of a contracting couple’s fertilized egg being implanted into the surrogate. The fetus is, therefore, not genetically related to the surrogate. Even if one ignores the value assigned to the gestational bond, it is critical to recognize that the surrogate is the patient while pregnant. Consent to any procedure, exam, or test, including abortion, requires the consent of the surrogate regardless of contractual language. Commercial surrogacy is risky for both parents and surrogates because, throughout the gestational period, surrogates and parents may disagree on some aspects of prenatal care, may change their minds, and may encounter a myriad of unanticipated issues, even with the strongest preimplantation contract. This case highlights many of the clinical, legal, and bioethical challenges involved with surrogacy. We, however, want to highlight specifically that the surrogate is the patient and thus, regardless of a contract with the biological parents, consent is required for any procedure or surveillance of the fetus during a surrogate pregnancy.

Ethical considerations regarding prenatal diagnosis of HLHS

McKenna West, the surrogate in this case, traveled to Texas from Alaska after the 20-week prenatal ultrasound in order to avoid the abortion the parents requested and to seek treatment for the fetus once born. The fetus had been diagnosed with hypoplastic left heart syndrome (HLHS), a serious congenital condition where the heart only has one functional ventricle. There is no cure, but a series of major cardiac surgeries can prolong life. These interventions can extend life years, but surviving patients have limitations in quality of life, and often still require a heart transplant or die. Survival statistics vary by study, but around 70% who undergo interventions or heart transplants are alive at 5 years of age.

While some children survive, treatment comes with significant burdens. Thus, there are value differences around the proportion of benefit to burden related to the arduous interventions after birth, each carrying risks of premature death. In general, it is up to parental discretion whether to terminate the pregnancy, continue the pregnancy with perinatal hospice, or seek post-natal interventions. 

Abortion, surrogate contracts, and ethics

Terminations anywhere are only ethically permissible where the patient, i.e., the pregnant person, requests and consents to the procedure. No clinician in any state should perform a termination against a patient’s wishes, even where it might be court ordered. However, surrogacy laws and varying legal views on what constitutes parentage may differ between states and jurisdictions. 

Contracts in commercial surrogacy are limited by the primary moral status of the surrogate as patient. While a fetus, parents have the ethical authority to decide what they feel is best. Clinicians, ethically, owe obligations to the patient. As such, clinicians involve the biological parents of the fetus only as much as the surrogate desires. This is because, as a patient, the surrogate has a right to privacy. The contract may include agreements to medical surveillance, testing, and nutrition supplements, but these still require the clinician to obtain the consent of the surrogate, whether the biological parents agree or not.

Once the child is born, they have their own moral status, and thus, in terms of treatment for the child, parents guide decisions. The question then is, who are the rightful parents with authority to represent the child’s interests and make value-based decisions towards care? 

Ethical considerations around parental authority

The issue of parentage in surrogacy is complex. It involves considerations of gestational time, genetics, and contractual agreements. In traditional pregnancies, the person who gestates the fetus and gives birth to the child is a genetic and legal parent. Technology, however, has changed this. Pregnancies can now be carried without a genetic connection, which puts parentage into the legal arena. Adoption of embryos, IVF, and surrogacy require legal processes to determine parentage. In this case, there are biological parents.

In HLHS, equally loving parents may make entirely different decisions. Some loving parents may feel that any chance at any type of life justifies the most extraordinary risks and interventions, e.g., a Norwood procedure. Other loving parents may feel that the extraordinary treatments required to extend life in HLHS, and the risks involved, are out of proportion to any of the potential benefits. This is the question that requires answering for this child. Whose values should speak for this child who cannot speak for itself- the parent who carried it through gestation or the biological parents? Ethically, the genetic parents of the child have a weightier claim, especially given their contractual agreement with the surrogate. This is because parents are determined, all things considered, to be the best parties to represent the values of their children. This should not be any different here, even with due consideration given to the emotional bond of gestation.

The Texas Case

While this case is grabbing headlines, the ethical issues around surrogacy in fetal health centers and clinical spaces are not new, though they generally don’t lead to conflicts. There is a lot of room for misinterpretation of this case. The surrogate is the patient and consent is required for any procedure or surveillance of the fetus during a surrogate pregnancy. This is just as true in Alaska as it is in Texas. Surrogacy contracts must not contain language that encroaches upon or imposes penalties for the surrogate’s exercise of their rights as a patient to bodily autonomy. Healthcare practitioners alike should never show deference to a surrogacy contract when treating a pregnant surrogate. It is important to note that the legal strategy pursued by Texas Attorney General Paxton raises significant ethical concerns. Paxton assisted in obtaining a court order through strategic use of the normally neutral child-protection system to usurp parental decision-making and legally order clinicians to provide stabilizing “life-saving medical care.” This limits values-based parental decision-making in a condition where there are essentially three recognized ethically and medically appropriate options for parents to choose (Norwood procedure, heart transplantation, or comfort care). By casting this case in a context of parental neglect, if the comfort care approach is chosen, it serves as a legal end-run, with the state invoking parens patriae to mandate a specific class of intervention in a condition where (an)other recognized option(s) exist. If this holds legally, this will be a significant encroachment on parental autonomy.
It may be tempting to see this case simply as a matter of life or death, but life through intervention here brings a significant amount of burden that any patient would have the right to refuse. For children, parents exercise autonomy based on their application of values to the medical facts. Therefore, the ethical question in this case is who is the best person to make value-based determinations of the child’s best interests, not which option is the best. We believe that any ethical analysis would determine the genetic parents as the most appropriate to make these value-laden decisions, especially given consideration to the previous freely agreed-upon contract. However, this determination has also now been cast as a legal question that a court will have to decide.

Ian D. Wolfe, PhD, RN, HEC-C is the Director of ethics at Children’s Minnesota
Rafael Escandon, DrPH, PhD, HEC-C, is the founder of DGBI Clinical Research and Ethics Consulting 

The post In The News…A Surrogate, Parents, and Politics: Clearing up the Ethics of Surrogacy Disputes appeared first on Bioethics Today.

This editorial appears in the August Issue of the American Journal of Bioethics

Epistemic humility is an important virtue for medical practitioners to show within clinical practice. Broadly construed, epistemic humility denotes a disposition to acknowledge the limitations of one’s knowledge and the corresponding willingness to share those limitations with others. Crucially, epistemic humility does not mean unwarrantedly downplaying one’s expertise or displaying false modesty but rather requires engaging with epistemic uncertainty. Within medical practice, this involves recognizing the fallibility of one’s judgments, remaining open to complementary perspectives, and fostering trust while guarding against misplaced certainty. Showing epistemic humility is also known to improve research and teamwork by encouraging responsiveness to evidence, integrating diverse knowledge traditions, and supporting skills like active listening, which help avoid epistemic injustices and enable “epistemic fluency”.

As these strands of work highlight, epistemic humility is not merely an individual attitude, but a key component of effective professional collaborations in medicine. As physicians are increasingly seen as “collaborating” with medical Artificial Intelligence (AI) systems, it is vital to consider how emerging socio-technical systems in medicine impact and could foster forms of epistemic humility conducive to clinical collaborations.

Human–AI Collaboration and Epistemic Humility

Collaboration between physicians and AI systems is widely endorsed as the most effective way to realize the potential benefits of medical AI. Human-AI collaboration is expected to offer advantages over relying solely on the AI system or solely on the human physician. Benefits include the augmentation of human skills, higher efficacy or efficiency than physicians alone and better patient outcomes. Nevertheless, some have warned against the issues of overreliance, automation bias and increasing dependence of physicians on AI systems, which may hamper effective human-AI collaboration. An authoritative presentation of algorithmic outputs may, for example, reinforce an unwarranted sense of certainty in physicians, who, without sufficient awareness of their own and the AI’s limitations, might be inclined to follow these outputs without adequately discussing relevant considerations with the patient.

In the academic literature, it is increasingly recognized that successful human–AI collaboration in medicine depends on factors that extend beyond considerations of efficiency and accuracy and are closely linked to key virtues and skills that guide medical practices. For instance, recently, it has been argued that AI needs to be designed to foster epistemic humility to know how to collaborate with these systems. As epistemic humility has long been regarded as an important intellectual virtue for human physicians and as a defining feature of the claims they advance, it is likely that AI systems that positively relate to this virtue are better adapted to medical practice. Yet, it is essential to avoid misleadingly anthropomorphizing AI systems as collaborative partners that possess certain virtues themselves. Given that AI systems cannot have knowledge, consciousness, self-awareness, and moral agency, they cannot possess the motivational and dispositional qualities that define virtues in human agents.

Technologies themselves are not moral agents, yet they can be designed to support, scaffold, or stimulate virtuous behaviors in human users. In Shannon Vallor’s account, technologies can function as virtue-conducive artifacts when they are structured in ways that encourage or enable humans to exercise those virtues. Building on her account, AI systems in medical contexts can be designed to “display” epistemic humility in ways that scaffold virtuous action in human collaborators. For example, an AI system might provide output that challenges physicians to justify their judgments, confronts them with recent empirical research and clinical guidelines, or notifies care providers if there are missing data from an individual patient to justify a specific course of action. Such design features could function as virtue-conducive affordances: they would prompt physicians to acknowledge the limits of their own knowledge, consider alternative perspectives, and engage actively with patient experiences and multidisciplinary input.

Depicting AI as displaying epistemic humility can be further substantiated by Rosalind Hursthouse’s action-oriented account of virtues. Her account suggests that virtues can provide action guidance, as virtues are not judged by the dispositions but by the actions one takes. This action-oriented account of virtue opens the conceptual space for attributing virtue-consistent patterns of behavior to entities that lack consciousness or moral agency, such as AI systems. Rather than asking whether such systems can possess epistemic virtues, the more pertinent question is whether they can be designed to instantiate or enact patterns of epistemic interaction that are consistent with those virtues. A medical AI system that systematically refrains from epistemic overreach, by avoiding unwarranted claims to authority or completeness, can thus be understood as epistemically humble in a derivative yet ethically meaningful sense.

On this view, epistemic humility functions as a collaborative and practice-oriented property that can be assessed based on externally observable actions rather than an intrinsic feature of the system. An AI system promoting or displaying epistemic humility does not merely generate accurate outputs; it structures epistemic relationships in ways that preserve the appropriate distribution of authority among clinicians, patients, and technological tools. By foregrounding the provisional and defeasible character of its contributions, such a system helps sustain the normative priority of human judgment and deliberation within clinical decision-making, without presupposing that human agents are epistemically infallible.

From the perspective of virtue-conducive design, AI systems can induce human virtues by creating virtue-conducive environments. They can be structured to prompt reflection on uncertainty, limits, and contextual factors that resist full formalization. Instead of presenting outputs as decisive and final answers, such systems should invite deliberation. Potential means of achieving this aim could be found in foregrounding uncertainty, in demanding justifications from clinicians, or in drawing attention to patient-centred considerations based on their experiential knowledge. Such “Socratic” features could help support, not diminish, clinical expertise, by helping clinicians calibrate appropriate epistemic self-trust, resist overconfidence, and remain open to competing considerations.

At the same time, the outputs of AI systems themselves should equally display epistemic humility. In an action-oriented sense of virtues, this would require AI outputs to reveal their epistemic limits by exposing the uncertainty of their predictions, their dependence on particular, and possibly limited, data sources, and the absence of relevant experiential information to avoid social misattributions. Especially in high-stakes medical contexts, epistemic humility would therefore not merely be an attractive design feature, but arguably a normative requirement for ensuring that AI contributes to, rather than destabilizes, the quality of clinical judgment. If AI systems are designed to consistently flag their limitations, they could well help to establish a collaborative environment in which epistemic authority is neither ceded to the system nor human expertise insulated from critique. Epistemic humility should therefore be considered a foundational design and interactional norm for systems intended for human-AI collaboration in medicine.

In practice, taking epistemic humility seriously when developing or using collaborative AI systems requires articulating how this virtue may be operationalized in medical care. In some instances, it may be sufficient for AI systems to display uncertainty through calibrated probability distributions or uncertainty quantification (UQ). This might be especially fitting in complex diagnostic or therapeutic settings, where clinicians must interpret available evidence to determine what ails a patient and which treatments seem most recommendable. By making uncertainty explicit, AI systems could prompt more cautious clinical reasoning here, reducing the risk of premature or overconfident conclusions.

Although this approach to medical uncertainty and the promotion of epistemic humility may be effective in contexts that can rely on ample biomedical evidence, other clinical situations may require an explicit signaling of epistemic boundaries. For example, displaying such a “scope boundary” could be needed when a medical recommendation also requires input based on a patient’s experiential knowledge and testimony. This boundary would indicate that experiential knowledge is missing from the system, emphasizing the importance of shared decision-making with the patient. an approach that is also in line with calls to guard against AI systems overreaching their warranted authority in treatment decisions.

Developing for Epistemic Humility

Fostering epistemic humility in and through medical AI could be a fruitful avenue for achieving the hoped-for benefits of human–AI collaboration. Epistemic humility also implies that medical AI intended for collaborative settings should avoid forms of algorithmic assertiveness that present outputs as authoritative and final, and instead include a designed posture of epistemic modesty that preserves the central role of human deliberation in clinical care.

Developing medical AI for epistemic humility requires careful engagement with how AI systems shape and affect humility at the human level. At the level of individual users, AI systems should be designed to promote reflective and self-critical decision-making, for instance by prompting users to engage more explicitly with their own line of thinking through “pointed questions” or uncertainty measures instead of clear-cut answers. Such design features can also support approaching human-AI disagreement as a signal of uncertainty rather than error, encouraging careful reassessment of assumptions on both sides. Beyond the individual level, it is also crucial to consider how clinician-AI collaboration reshapes epistemic humility in human-human collaborations more broadly and investigate how AI can be designed to support valuing the perspectives, distinct knowledge, experiences and needs of others. AI is, for example, envisioned to assist collaboration between different healthcare workers and geographically disparate healthcare teams. As Cajas Ordóñez et al. argue, there is an opportunity here to foster epistemic humility, as “humility also extends to interprofessional collaboration, where AI systems can serve as tools for democratizing clinical knowledge and supporting team-based care.” Building on this, epistemic humility should be understood as a leading design feature of AI systems that can contribute to, rather than threaten, the ability of human users to engage with the limitations of their knowledge and different epistemic perspectives.

In sum, while epistemic humility is a normatively attractive ideal for medical AI, its ethical and clinical value depends crucially on an appropriate pairing with epistemic competence, in what might be described as epistemic calibration within clinical practice. Diagnostic or prognostic outputs that are not proportionate to the level of medical evidence currently available cannot, even if cautiously framed, meaningfully support patient care, while overly cautious AI output in cases of very high certainty may prove equally misleading. This is particularly important in high-stakes medical settings, where both overreliance on automated outputs and their systematic dismissal can lead to harm. Epistemic humility, when grounded in demonstrable clinical performance, can help align the perceived authority of AI systems with their actual epistemic merits, supporting more appropriately calibrated trust and more reliable collaborative clinical decision-making.

Georg Starke, MD, PhD, Jojanneke Drogt, PhD, and Karin Jongsma, PhD

The post Computer Says: I Don’t Know? – On Epistemic Humility as a Condition for Human-AI Collaboration appeared first on Bioethics Today.

This editorial appears in the August Issue of the American Journal of Bioethics

The ethical discourse surrounding artificial intelligence (AI) in healthcare has largely focused on algorithmic performance, bias, privacy, transparency, and explainability. These concerns remain critically important. However, as AI applications increasingly move from development environments into clinical and research operations involving patient health and medical outcomes, a complementary ethical challenge is emerging: how AI changes the workflows through which decisions are made. I argue that ethical evaluation must therefore expand beyond algorithmic performance to include what might be termed workflow integration ethics, the study of how AI influences decisions, behaviors, and outcomes within healthcare delivery systems.

The articles in this issue by Rentzepis et al., Char et al., and Hatherley et al. each examine different AI applications: research recruitment, clinical summarization, and federated learning. Yet collectively they reveal a broader concern. The ethical risks associated with AI frequently arise not from the model itself, but from the interaction between the model, the humans using it, and the organizational systems into which it is deployed. Healthcare organizations often evaluate AI as though it were a standalone technology. In practice, AI functions as an embedded participant in complex sociotechnical systems with pre-established operating goals, workflows, and accountability structures. Healthcare AI is not deployed into a vacuum; it is deployed into systems and practices of healthcare delivery. Ethical assessment must therefore extend beyond evaluating algorithms to evaluating their role and performance within care delivery workflows. As a result, the ethical unit of analysis must expand from algorithms to the sociotechnical systems in which they operate.

From Algorithm Ethics to Workflow Ethics

The ethical concerns raised by AI recruitment tools are often framed as questions of fairness, representativeness, and privacy. An AI recruitment model may alter which patients are approached about clinical trial participation. Similarly, concerns about generative AI summarization focus on accuracy, hallucinations, and clinician trust. A summarization tool may shape how clinicians understand a patient’s history. Federated learning raises questions regarding transparency, accountability, and data governance. A federated learning model may influence predictions despite uncertainty regarding the data on which it was trained.

While these concerns appear distinct, they share a common characteristic: they question how information moves through healthcare delivery systems and ultimately influences human decisions and the experience and health outcomes of patients affected. In each case, the ethical concern is not solely whether the model functions correctly. The concern is whether healthcare organizations can understand, monitor, and govern the workflow consequences that follow from AI-generated outputs.

AI as A Workflow Intervention

Healthcare organizations routinely evaluate interventions that alter clinical workflows. New staffing models, order sets, triage protocols, and quality-improvement initiatives are all assessed according to their effects on care delivery processes and patient outcomes. AI should be treated similarly. An AI tool rarely acts independently. Instead, it changes the timing, sequence, prioritization, or content of decisions made by humans making judgements and performing tasks. These workflow effects may ultimately be more consequential than the model’s computational performance characteristics. For example, an AI recruitment tool that identifies eligible patients more efficiently may increase trial enrollment. However, it may also alter who is approached, when they are approached, and how recruitment resources are allocated across populations. That influence on enrollment can alter the outcomes of the research study. Similarly, a clinical summarization tool may produce technically accurate summaries while subtly changing clinician attention, documentation practices, or information-seeking behaviors in ways that change the quality or nature of care received by their patients. The ethical question therefore becomes how AI reshapes human decision-making within operational systems and whether healthcare organizations can responsibly balance the benefits of those changes against their consequences

Accountability Requires Observability

The articles by Char et al. and Hatherley et al. appropriately highlight concerns regarding transparency and performance visibility. This is not merely a technical problem; it is also an operational and governance problem. Healthcare organizations cannot govern what they cannot observe. Historically, clinical governance has depended upon the ability to reconstruct decision pathways, identify failures, and implement corrective actions. AI introduces new layers of complexity into these processes. Recommendations may be generated external to the organization while being shared within it, training data may be inaccessible, and model outputs may influence workflows in ways that are difficult to detect retrospectively and therefore require prospective evaluation. As a result, healthcare organizations require mechanisms that enable ongoing observation of AI behavior in operational settings. Ethical oversight should not end at deployment. It should include prospective monitoring of workflow effects, user interactions, simulated versus real-world performance, overrides, functional shift and drift, unintended consequences, and differential impacts across patient populations. This requirement is particularly important because many harms emerge only after implementation.

The Importance of human-AI Teaming

A recurring assumption within healthcare AI discussions is that ethical concerns can be addressed by maintaining a “human in the loop.” While human oversight remains important, simply inserting a human reviewer may be insufficient. The relevant question is not whether a human remains involved, but whether the human-AI team functions effectively. In many healthcare settings, the more accurate description is not a human-in-the-loop system, but a human workflow with “AI in the loop.”

Research across healthcare operations demonstrates that outcomes depend upon communication structures, role clarity, feedback mechanisms, and organizational culture. Similar principles should guide AI implementation where AI functions as a member of a care delivery team rather than an independent agent. Organizations should therefore evaluate whether: 1) users understand AI outputs, 2) workflows sequence and objectives are understood well enough to permit meaningful review, 3) accountability for action and workflow outcomes remain clear, 4) users are empowered to override, adjust, or reevaluate recommendations, 5) feedback loops exist to improve performance. These factors influence safety and effectiveness as much as algorithmic accuracy. In this regard, the integration of AI into healthcare workflows may have more in common with the implementation of medical devices and pharmaceuticals than is often acknowledged. The technology itself matters, but so do the systems, training, governance structures, and human behaviors that determine its real-world impact.

Ethical Success Requires Implementation Science

The next generation of healthcare AI ethics should incorporate principles from implementation science, health services research, and systems engineering. Healthcare has repeatedly demonstrated that interventions with strong efficacy can fail when implemented poorly. Conversely, interventions with modest technical advantages may achieve substantial impact when integrated effectively into clinical operations. Ethical evaluation should therefore examine not only whether an AI model works, but whether healthcare organizations possess the governance structures necessary to deploy it responsibly. This shift has important implications for regulators, health systems, sponsors, and institutional review boards. Questions regarding performance, bias, and privacy remain necessary. However, they should be accompanied by questions regarding workflow integration, organizational accountability, monitoring plans, and mechanisms for continuous learning. Ethical success will depend not only on what an AI model does, but on how healthcare organizations choose to implement, monitor, and govern it.

Conclusion

The papers by Rentzepis et al., Char et al., and Hatherley et al. collectively illustrate a transition occurring across healthcare AI. The central ethical challenge is no longer simply evaluating algorithms. It is understanding how AI becomes embedded within the workflows through which healthcare and research are conducted. As AI moves from experimentation to operational deployment, ethical oversight must expand accordingly. The future of ethical AI in healthcare will depend not only on building better models, but also on building more accountable systems in which those models operate. Healthcare organizations should therefore evaluate AI in the same way they evaluate any other intervention intended to improve care delivery; not by what it predicts or computes alone, but by how it changes real-world activities, decisions, and outcomes.

Maame Yaa A. B. Yiadom, MD, MPH

The post The Ethics of Integration: Why Healthcare AI Must Be Evaluated Within Clinical and Research Workflows appeared first on Bioethics Today.

This editorial appears in the August 2026 Issue of the American Journal of Bioethics

Artificial intelligence (AI) agents are being used to support a range of healthcare activities, offloading tasks that have traditionally been performed by human workers. These tools have a variety of use cases in healthcare and currently are being used to support medical decision-making and diagnosis, scientific discovery, clinical documentation, appointment scheduling, and patient triage. Multiple societal and institutional pressures have highlighted the importance of reducing administrative costs in healthcare, making AI agents an attractive solution for generating efficiencies. While there are mixed opinions on whether these tools will deliver on their promises related to efficiency gains and cost reductions, their potential has generated major investments in applications of healthcare-related AI.

This issue of AJOB highlights several contexts in which AI agents are likely to play a significant role in healthcare. Char et al. focus on AI-enabled clinical documentation, an area where AI agents may support clinical note writing and other documentation-related tasks with greater efficiency. Rentzepis et al. highlight other applications of AI in clinical research, where AI agents may support subject recruitment, clinical trial design, data analysis, and study management. It is encouraging to see bioethicists examining these and other “early-use cases” involving AI agents, adding to an ever-growing literature exploring ethical dimensions of healthcare AI.

We anticipate that this trend will continue as the number of AI applications in healthcare expands. The work of bioethicists will be crucial for understanding the moral complexities of these technologies, particularly in contexts where AI tools are tasked with autonomously performing the work of healthcare professionals.

What is less clear is the extent to which AI agents will impact the day-to-day work of bioethicists. As health systems look for ways to extend resources and create workflow efficiencies it is likely that pressures to expand the use of AI agents in bioethics-related work will result. We suggest that bioethicists ought to anticipate this possibility and be proactive in considering how AI agents may redefine their work. In this brief editorial, we consider several potential applications of AI agents in bioethics, highlighting how these tools might impact activities in clinical ethics, bioethics research, and ethics education.

There are already several opportunities for clinical ethicists to integrate AI agents into their work. For instance, AI agents might be used to extract and summarize information from a patient’s electronic health record or to identify laws, regulations, and institutional policies that may be relevant to an ethics consultation. Similarly, AI agents might be used to support clinical documentation activities done by clinical ethicists, for example, drafting clinical notes based on a family meeting or a complex care discussion. Additionally, conversational agents might be used to gather stakeholder narratives or to clarify patient values.

In addition to these AI tools, emerging agentic AI systems are being developed to support ethical analysis and clinical decision-making. For instance, Dutta Roy has presented a Bioethics Artificial Intelligence Advisory (BAIA) framework in which multiple ethical frameworks are collaboratively applied to cases in clinical ethics. In the future, systems like the BAIA might be used by clinical ethicists to confirm a clinical recommendation, providing a second opinion on their moral reasoning or analysis of a complex case. It is also possible that future AI systems might augment the moral reasoning capabilities of healthcare professionals who have not had formal training in bioethics, which may be particularly beneficial in resource-limited settings where a formal ethics consultation service is not available.

In a very different setting, bioethics researchers might also use AI agents in support of their work. For instance, AI agents have been developed to support literature reviews and synthesis of large datasets. Of note for bioethics researchers, there are emerging agentic AI systems that aim to simulate or automate aspects of qualitative research. For instance, AI systems are being developed to simulate focus-group discussions, which can include pre-defined participant archetypes or personalities. Other AI agents may autonomously conduct qualitative interviews or moderate focus groups. While these are nascent technologies, they could significantly reduce both the cost and time required to conduct qualitative bioethics research.

Lastly, bioethics educators also have a multitude of opportunities to integrate AI agents into their teaching. Tutoring-focused agents might aim to create personalized and adaptive guidance for ethics learners, particularly in contexts where ethics education is completed asynchronously. Other AI agents may support instructor-led simulation activities as a means of discussing complex ethical cases and fostering moral deliberation amongst students.

Beyond serving as support tools for bioethics educators, future agentic AI systems might serve as ethics instructors in a stronger capacity. In a previous volume of AJOB, Rahimzadeh and colleagues called attention to potential uses of generative AI tools in the teaching of bioethics, noting biases in the ethical frameworks reflected in responses. Emerging agentic AI systems may help to address issues of bias in ethics education by incorporating multiple ethical frameworks in their design and prioritizing ethical reflection over conflict resolution. This is not to say that these systems are comparable to instructor-led teaching, but they may offer an alternative approach to some types of ethics education.

Our sense is that in comparison to other healthcare professionals, many of whom are more proactively exploring the use of AI tools in their work, bioethicists have tended to be a bit more skeptical about the potential promise of AI agents. As a result, bioethicists may want to consider the experiences of professionals in these other fields, where early AI adoption has begun, such as radiology, and consider analogous strategies for the using AI agents in our work.

By highlighting several areas where AI agents may impact the work of bioethics, we aim to call attention to the importance of reflecting on the scope of acceptable and unacceptable uses of these tools in bioethics-related activities. A wide range of AI agents will likely be accessible to bioethicists soon, and it will be incumbent on individual practitioners to decide for themselves which of these tools are appropriate for their work. Whether it be to support a care team in a complex ethics consultation or to enhance the delivery of ethics-related content in a classroom, bioethicists will have many opportunities to use AI agents in their work.

Unfortunately, there are still many uncertainties related to these emerging AI agents. For instance, will a documentation tool capture subtle points of consensus or disagreement in a complex clinical discussion related to goals of care? Will an AI agent perform as well as a trained research analyst in collecting and analyzing qualitative data? Will an educational agent recognize the unique needs of bioethics learners and respond accordingly? By proactively asking such questions, and potentially contributing to the development and evaluation of AI agents, bioethicists can be better prepared for the many changes that lie ahead as healthcare systems deploy new AI tools. In this regard, we suggest that bioethicists should not only examine the ethical complexities of AI agents being deployed in other areas of healthcare but should consider the potential impact of these tools on the work of bioethics itself.

Notes

We use the term “AI agent” to refer to a generative AI system that autonomously executes multi-step tasks with limited human oversight.

Austin M. Stroud, MA & Richard R. Sharp, PhD

The post Reimagining Bioethics in the Era of AI Agents appeared first on Bioethics Today.

While looking into credentials in clinical ethics, I came across one I had never seen before: the “Certified Health Care Ethics Consultant,” with the initials HCEC-C. For a moment, I thought I was looking at a typo. The credential I know well, the one I hold, is the Healthcare Ethics Consultant-Certified, or HEC-C. This was something else.

The HEC-C credential dates to 2018 and is administered by the HCEC Certification Commission with the support of the American Society for Bioethics and Humanities. More recently, the American Institute of Health Care Professionals (AIHCP), a private continuing-education provider, began offering the HCEC-C credential. It is a real credential, offered by a real organization, in a growing market of professional credentials.

The problem is obvious: the titles are nearly identical, the initials are almost the same, and HCEC also names the HCEC Certification Commission. A hospital deciding who should perform ethics consultations, a clinician reading a consult note, or a patient viewing a consultant’s profile could easily mistake HCEC-C for HEC-C.

The ethical problem is not that two organizations offer different credentials for ethics consultants. It is that nearly identical names can mislead readers and obscure substantial differences in what the credentials require and demonstrate. Employers and prospective certificants, therefore, need to judge what stands behind a credential, not what its name and initials seem to imply.

Why this confusion matters

Credential confusion in clinical ethics is not just an academic annoyance. It affects how hospitals decide who should perform ethics consultations and how patients and families understand who is advising them on some of the most difficult decisions in their care.

Confusion of this kind is part of a broader problem. Across professions, the proliferation of certificates, badges, and other credentials has created what researchers describe as a “maze” that confuses consumers, employers, and even educators. Research on non-degree credentials shows that, in a crowded credential market, quality signals can be weak: names and initials may convey apparent legitimacy even when the underlying requirements differ sharply. Employers and prospective candidates may have little practical ability to distinguish rigorous professional certifications from credentials based principally on completion of educational courses.

The HCEC-C/HEC-C overlap illustrates that problem. The two credentials differ in eligibility, education and experience requirements, assessment, and renewal. Hospitals should not have to untangle those differences merely to determine what a credential signifies. A separate detailed side-by-side comparison is available for readers who want to see how different the two credentials actually are.

How a credential can be judged

The field cannot prevent every confusing credential from appearing, but several basic features of a certification program can be evaluated without specialized knowledge of clinical ethics. A central purpose of professional certification is to protect the public by distinguishing people who are competent to perform a defined role. Standards for how a certification program should be built exist for that reason, and they were developed outside the bioethics field. Widely recognized benchmarks include the NCCA accreditation standards and ANSI/ISO/IEC 17024. A decade ago, I identified these standards as a roadmap for developing a high-quality certification process for clinical ethics consultants in an American Journal of Bioethics article.

This table lists several core requirements from those standards:

Entries reflect public information on each organization’s website as of August 2, 2026. Sources:  HEC-C Initial Examination Candidate Handbook; HEC-C Renewal Handbook; NCCA accreditation record; AIHCP Health Care Ethics Consultant Certification; AIHCP Health Care Ethics Consulting Certification Program Information; AIHCP Health Care Ethics Consultant Recertification; and AIHCP Advisory Boards.

Why there is no easy fix

When a problem like this appears, the instinct is to look for an enforcement mechanism: a cease-and-desist letter, a regulatory complaint, or a rule about who can use which initials. But clinical ethics consultants generally are not licensed as such, and no dedicated licensing board regulates who may use these credentials. Employers and patients, therefore, cannot turn to a profession-specific regulator to determine what a credential signifies or whether its use is misleading.

ASBH holds three live registered trademarks in this area: HEC-C, HCEC Certification Commission, and Healthcare Ethics Consultant-Certified. I am not a lawyer and will not comment on whether the similarities raise legal questions. Regardless, trademark law cannot prevent all credential confusion: many similar names will not infringe anything, and litigation is slow, costly, and uncertain.

Who decides in practice?

In practice, employers often determine which competing credentials carry weight. CPR certification provides a familiar example. Several reputable organizations offer programs, while a large online market in cheaper cards has also emerged. Hospitals commonly respond by specifying which organizations’ cards they will accept.

Candidates make choices, too, and they have apparently treated the two credentials quite differently. Although information about HCEC-C had been available online since at least 2024, the August 2, 2026 comparison identified only one person in AIHCP’s public opt-in directory, compared with 1,010 in HEC-C’s.

Government agencies also distinguish among training programs and credentials. For example, Florida publishes a Master Credentials List that applies a formal Framework of Quality. I am not aware of a comparable review of clinical ethics credentials, but employers, candidates, and regulators are all making choices that collectively determine which credentials are recognized and used in practice.

Credential confusion in clinical ethics will not disappear because new credentials will no doubt continue to appear. Standards for certification programs are public, however, and criteria such as those listed in the table above can be applied to any certification program in any field. Hospitals should therefore specify which certifications they recognize for particular consultation roles, verify credentials with issuing organizations, and examine what each credential requires. Prospective certificants should do the same before investing their time and money. In clinical ethics, one letter may be the only visible difference between two credentials, but it should not be the basis on which either is judged.

Ellen Fox, MD, HEC-C, is President of Fox Ethics Consulting, Arlington, Virginia.

The post One Letter Apart: Credential Confusion in Clinical Ethics appeared first on Bioethics Today.

The All of Us research biobank recently announced a massive new data release – making it the largest health database in the world that allows researchers to access both donated genomic information as well as participants’ electronic medical records (EMRs). That combination can allow researchers to associate the genomic variations that we all have with health outcomes (e.g., developing cancer) – while also controlling things that can impact our health, like behavior (e.g., smoking) or environment (e.g., living in a city polluted with smog). These kinds of non-biological factors can protect us from or accelerate the impact of genomic variation, making them critical to understand for research.

While the biobank has existed since 2018, the enhanced ability to connect donated samples and genomic sequencing with EMR data is new. Previously, the biobank mostly relied on participants to connect and share their own EMR, or participating hospitals agreeing to share it on their behalf. It is now utilizing the eHealth Exchange health information network  for clinical data sharing to facilitate the process.

What makes All of Us particularly special, and therefore this news even more impactful on genomic science, is its focus on recruiting participants historically underrepresented in research. Roughly 80% of All of Us participants are from such communities. This is in stark contrast to global genomic databases, where almost 80% of participants in genome-wide association studies are of European descent. Even in government-supported databases, like the NHGRI-EBI GWAS Catalogue, only 2.4% of participants are of African ancestry.

A lack of diversity in genomic research is not just bad for communities underrepresented in these databases – it is bad for everyone. Yes, as a bioethicist, I would of course argue that it is bad to live in an inequitable society at the existential level, but as a scientist, I also know that the lack of data diversity can limit health advances for everyone. For example, the high genetic variation within populations of African ancestry can make participants uniquely situated to contribute to scientific advancements. That 2.4% of participants with African ancestry in the NHGRI-EBI GWAS catalogue have contributed to 7% of findings regarding association of genetic variation with outcomes.

Representation in genomic research also matters because not only can different genomic variations sometimes be associated with ancestry, but the impact of health disparities can also be associated with self-identified race and ethnicity due to factors including institutionalized racism. But to be able to identify improved health outcomes generalizable to, or targeted toward, historically excluded research populations, researchers need access to representative data.

As law and ethics faculty at the University of Michigan Medical School, my work focuses on the use of health data for medical research and ways to increase access. I was also the principal investigator of a 5-year grant from NHGRI focusing on how academic genetic researchers chose data for their work, and how to improve the system to encourage and support the use of representative health data. What we found is critical to fully understanding the potential impact of All of Us.

When we interviewed genetic researchers and asked them how they chose databases for their work, they all talked about the importance of large numbers of participants, but not one brought up looking for genomic diversity. Although when pressed, most agreed it was important. But when we surveyed hundreds of researchers, we found that the vast majority reported wanting to work with data from non-European ancestries.

But why would researchers report that they want to work with representative data and admit that they don’t bother to look for them? Most likely because representative data do not generally exist in the first place.

It’s very time-consuming  and complicated for researchers to collect their own data from participants from scratch, meaning, in some cases, actually walking from patient waiting room to waiting room asking people to spit into a tube for research. Doing that will also just reflect the differences in demographics of people who are sitting in that waiting room (including that they likely have insurance) – overall often making it a painful, and not always fruitful, process.

Another possible solution is for researchers to pull data from several different databases to combine non-Euro-centric data. But here they can also face expensive and time-consuming tasks related to “harmonizing” the data to standardize health comparisons. In addition, many top-tier journals are more likely to publish findings from larger databases. If researchers feel pressure to work quickly and publish in high-impact journals for promotion (which, it turns out, they do), they are more likely to turn to existing databases. If they turn to existing databases, their research can only be as representative as the data already contained in them. You see how the problem becomes cyclical.

While some have argued that supporting researchers with different kinds of priorities would rectify these problems, we found that a lack of access to representative data was a systemic issue. We did not find an association between researcher demographics and ancestral populations represented in their publications. Data availability was consistently the major driver of use, meaning that data diversity cannot be resolved just by individual researchers alone. It is a collective action problem.

There remain challenges to using the All of Us biobank at scale. For example, just because the majority of data housed in All of Us are representative in comparison to those historically included in research, they are sometimes problematically different from each other. Genomic research often requires large numbers of homogeneous participants to compare variation and outcomes. Researchers might still have to resort to harmonizing data across different databases to have enough comparison points.

In addition, All of Us generally has a more comprehensive participant consent and community oversight process than other biobanks. Other studies have found that differences in consent rates to genomic research can be associated with race and ethnicity. The All of Us approach likely contributed to its ability to build a representative database in the first place. But specific consent preferences can also limit researchers’ ability to use and harmonize data to build larger cohorts if databases have different consent rules.

Overall, All of Us’ recent announcement is excellent news for genetic researchers and patients alike. But our research has found that challenges remain in translating representative data into representative research. Without dedicated investment in harmonization infrastructure to combine databases, even representative databases might fail to accomplish broader health goals.

This work was supported by the National Center for Advancing Translational Sciences (R01TR004244, UM1TR004404) and the National Human Genome Research Institute (K01HG010496). The author declares no conflicts of interest.

This work was supported by the National Center for Advancing Translational Sciences (R01TR004244, UM1TR004404) and the National Human Genome Research Institute (K01HG010496).

Kayte Spector-Bagdady, JD, MBe is the Wantz Professor of Bioethics at the University of Michigan Medical School

The post Even with New EMR Release from ‘All of US,’ the United States Still Hasn’t Solved its Genetic Data Diversity Problem appeared first on Bioethics Today.

When the doctor’s note from my daughter’s wellness visit to the hospital was posted on our patient portal, I discovered that the doctor’s AI notetaker had erroneously entered ‘alcohol’ as the ointment applied to a dry skin patch on my daughter’s face, when in fact, my response to the doctor’s question of what I applied to my daughter’s face was ‘Aquaphor.’ Not only did the AI notetaker replace Aquaphor with alcohol, but the scenario revealed that the doctor was probably not actively listening to my responses, perhaps because the AI notetaker used for recording had also been inadvertently delegated the task of active listening.

The integration of frontier Artificial Intelligence (AI) systems into healthcare and medical practices may have brought enhanced productivity and efficiency in the delivery of healthcare services; however, the focus of this essay is when the integration of AI into healthcare takes away more than it gives or leads to avoidable errors, like the replacement of “alcohol” with “Aquaphor” in a note.

AI is currently used by hospitals for management, administration, clinical decision-making, diagnosis, predicting patient outcomes, and personalizing treatment plans. Generative AI (GenAI) tools are used for various tasks, including generating clinical notes, recording inpatient visits, and summarizing telehealth transcripts. For example, Microsoft’s Dragon Ambient eXperience (DAX) copilot is a Generative AI tool used in healthcare. It was launched in 2023 and is used by close to 1,000 clinicians in the US, offering automated documentation of patient visits based on voice recordings, among other capabilities. Furthermore, Dax Copilot is integrated with Epic, a widely used electronic health record platform used by over 40% of hospitals in the US.

When emphasis is placed on efficiency optimization in a healthcare context, the question of what is gained and what is lost becomes important. Gen AI systems in healthcare  is increasingly mediating how clinicians capture and create patient-related documentation through the diverse capabilities and intuitiveness built into the AI systems. It follows that this integration of Gen AI and other frontier AI systems into healthcare will come with an eventual decline in a clinician’s dexterity and skills necessary for the optimal delivery of health care, an example of which is active listening, which is very important for doctor-patient encounters.

When AI notetakers become commonplace in clinical encounters, listening to the patient almost becomes the same as recording the patient (of course, after consent has been given), and when doctors delegate listening to the patient’s concerns to AI notetakers, the need to truly listen, actively and intently, to the concerns and responses of the patient reduces. When that happens, efficiency and productivity are unintentionally prioritized over quality of interaction, and that is where the challenge lies.

Doctor-patient encounters often involve nuanced, personal, and humane forms of person-to-person communication that can be lost on AI systems. There are ways in which words are said, with certain types of body language and non-verbal cues that will be almost impossible for AI systems to capture or decipher. When we increasingly allow AI systems to mediate such interactions, some things are gained and some things are lost. Efficiency and productivity are gained, while a genuine human-to-human connection is lost. Furthermore, possible overreliance on AI tools in the example with my daughter and I left the clinician in a place where the accuracy of the AI notetaker was not questioned. It doesn’t take a medical degree for anyone to critically question why a person would think of, let alone apply alcohol to the face of a 6-month-old baby, but that sort of concern can only be realized within a human-to-human conversational context. 

While using technology certainly has its benefits, it is very important that clinicians themselves take seriously their responsibility to review the summaries and notes produced by the AI tools, and that any errors should be caught at the review phase. My initial response to the error in the doctor’s note was to call out the doctor (and her assistants) for the error and ask how it was possible that such a mistake was not captured and corrected before the note was signed off and posted to our patient portal. But knowing what I know about AI and agentic systems, especially Gen AI and Large Language Models (LLMs), catching such errors is mostly easier said than done.

By design and functionality, Gen AI and LLMs that operate through AI note takers are built to be as convincing as possible, so that whatever is produced is almost believed to be accurate at face value. Gen AI systems have the tendency to sometimes generate and manufacture non-existent information and present it as factually correct to the unsuspecting user. A study conducted by a Boston University professor revealed that most workers who used agentic AI systems were less likely to trust their corrections of the questionable output produced by the AI systems.  While the study was conducted in a business work environment, when used day in, day out for several months or years in a clinical setting, clinicians are very likely to be less inclined to doubt the output produced by AI note-takers. Moreover, a skill like active listening is best developed and honed when it is put to constant use, and it is the grit developed from mastering such a skill over time that will most likely come together in developing the thoroughness and confidence needed to double-check the output of a seemingly efficient AI notetaker, especially when the AI tool has been praised for reliability and accuracy by its developers and investors.

Recently, there has been an increase in billion-dollar investments in deploying AI in healthcare, leading to greater interest and adoption of AI across many healthcare systems and exerting subtle pressure on clinicians to adapt to these technologies, some of whom may not even be able to decline. When technologically-driven adaptive skills are prioritized for clinicians to learn, other humane, personal, and non-technical skills will be unintentionally dismissed.

Therefore, rather than emphasizing technological adoption, more effort needs to be directed towards ensuring that the very skills necessary to provide excellent and humane patient care are not neglected. If doctors unknowingly delegate active listening to AI note takers, an overreliance on those AI systems over time will leave clinicians lagging behind in actively listening and eventually dulling foundational human communication skills. Nobody wants the services of a doctor who does not actively listen to their concerns or misses out on important details because the task of actively listening has been delegated to an AI tool. Over time, clinicians may naturally relent in honing their active listening and other humane skills.

To prevent this, a framework of collaboration should be developed. The efficiency and productivity advanced by AI in healthcare are best achieved when the human capabilities of clinicians are improved and supported. Clinicians should be encouraged to engage in activities that help hone human skills of effective listening, effective communication, and human-centered interactions. Efficiency and optimal productivity through AI must be balanced with human connection and humane quality of care, especially because many patients expect that human-to-human connection from their clinicians. Increased productivity should not be engineered at the expense of quality care within doctor-patient interactions.

 Blessing T. Adewuyi, PhD is an Instructor at the University of Georgia.

The post From Aquaphor to Alcohol: Can Overreliance on AI Lead to a Decline in Human Connection Within Healthcare? appeared first on Bioethics Today.

Kristen Biel had breast cancer and her Catholic school refused to renew her contract. Miriam Grussgot had a brain tumor and her Jewish school fired her. Both wished to seek recourse in the courts for alleged discrimination. Just a month ago, a Catholic school refused to renew the contract of a teacher who uses a wheelchair and immediately advertised for a person who can walk, and lawyers don’t want to take the case. Over 328,000 teachers work at the roughly 23,000 religiously affiliated schools in the United States. Why don’t teachers at religious schools have access to the courts for employment discrimination cases?

In 2020, the Supreme Court broadened the ministerial exception, a little-known loophole allowing religious institutions to discriminate against their employees despite generally applicable federal laws. The exception holds that to freely exercise their religion, religious institutions must be able to hire and fire “ministers” for any or no reason. The exception is overly broad in two ways: it bars all discrimination claims regardless of relevance to religious tenets; and it applies to secular lay employees whose primary job duties are not religious in nature. While the controlling case was decided several years ago, the chilling effects are now in full swing.

How we got here

In 2012, in Hosanna-Tabor Evangelical Lutheran Church and School v. EEOC, the Supreme Court applied a four-factor function test to the ministerial exception: religious title, e.g., minister; religious training; whether the employee held herself out to be a minister; and whether job duties included conveying the church’s religious message. In that case, the disabled teacher’s claim was precluded because of her deep religious training, her title, position, and duties.

In 2020, in Our Lady of Guadalupe School v. Morrissey-Berru, a lay teacher whose primary duties were teaching secular classes alleged discrimination. The Court considered a companion case, St. James School v. Biel at the same time. Morrissey-Berru claimed age discrimination and Biel claimed she was fired due to her breast cancer. The Court held the ministerial exception applied to the teachers despite their lack of ministerial title, minimal religious training, and secular duties. Therefore, all discrimination claims were barred before any discussion of the merits. Both Morrissey-Berru and Biel taught primary subjects including math, science, English, spelling, etc. and one also taught religion. Neither had extensive religious training.

Furthermore, the Court took the teeth out of its 2012 function test by suggesting churches are the only entities well-positioned to explain the ministerial function and that no strict formulas should apply. The potential wrongdoer now holds all the cards.

Biel eventually died of her cancer, unemployed. Justices Sotomayor, joined by Justice Ginsburg, began her dissent saying, “Two employers fired their employees allegedly because one had breast cancer and the other was elderly.” The dissent notes that the decision interprets the word “minister” so broadly that “it strips thousands of schoolteachers of their legal protections.” All sorts of employment discrimination cases, including disability, are now routinely dismissed throughout the US from district to circuit courts. When a religious institution wants to terminate employment or fail to renew a contract due to sicknesses like cancer or disabilities like an inability to walk unassisted, the church is free to do so legally. In most settings, religious exemptions from generally applicable laws apply to religious circumstances, i.e., the entity wanting an exception has a reason related to its ability to practice religion or hold a belief. Yet here, the teachers were not fired for actions that interfered with the religious institutions’ ability to practice their religions. Generally, having a disability or becoming old is not inconsistent with church doctrine or tenets.

Why It Matters

Those with disabilities face significant employment discrimination. Only 22 percent of people with disabilities are employed compared to 65 percent of those without. The firing and failing to renew contracts negatively impact not just the ability to earn money and important benefits like health insurance, but also the ability to participate in working life with its social, educational, and cognitive benefits.

As it stands, disabled teachers cannot feel secure in their employment at religious schools. While it is too late for Kristen Biel, the ministerial exception has spun out of control. Ethically speaking, the Americans with Disabilities Act should apply broadly and the ministerial exception narrowly, only to actual ministerial workers terminated for reasons relevant to religious tenets.

What to do

Left without access to the courts, teachers should approach accrediting organizations and challenge them to take a stand. Accreditors are not the government and do not owe any organization religious freedom to discriminate against disabled teachers. Another appropriate redress may be media attention to individual cases.

Free exercise of religion cases generally address permitting otherwise impermissible acts that religion requires (like allowing a religious incarcerated person to have a beard in conflict with a rule prohibiting one). There is no relationship between firing the disabled and the tenets of most religions; there certainly isn’t a religious calling to do so and not doing so would hardly seem to burden a religious institution in a religion-relevant way. For now, freedom of religion allows the religious to fire disabled teachers simply because they are disabled.

Anne Zimmerman, JD, MS is a Lecturer at Columbia University and Editor-in-Chief of Voices in Bioethics

The post Holy Immunity: The Ministerial Exception and Teacher Discrimination appeared first on Bioethics Today.

Bioethicists have become very good at arguing for transparency. We encourage institutions to disclose conflicts of interest, explain difficult decisions, communicate uncertainty, and justify ethical trade-offs. These are all necessary, even essential — yet they are not enough.

Transparency helps explain decisions. It does not, by itself, create the relationships required for public trust. Bioethics should therefore move beyond transparency toward a more demanding ethical practice: public enrolment.

Many of the most pressing challenges in contemporary bioethics depend on relationships that extend over years. Public health authorities ask communities to participate in vaccination campaigns, population screening programs, and epidemiological surveillance. Health systems increasingly rely on artificial intelligence to support diagnosis and resource allocation. Researchers ask citizens to contribute health data to improve care for future patients.

In every case, institutions are asking people not simply to accept a decision, but to remain engaged in an ongoing collective project. Transparency alone cannot sustain that engagement.

Explanation does not create a relationship

Being told why a decision was made, even when the explanation is transparent and offered in good faith, does not necessarily make people feel included in the process that produced it. Much of what passes for public engagement remains unidirectional: authorities speak, and the public listens. Feedback, when it is invited, is often symbolic — an opportunity to react rather than to influence outcomes.

A common response is to call for more participation. Yet participation itself is frequently treated as a procedural requirement: hold a consultation, invite comments, convene an advisory panel, and move on. Once the exercise ends, so too does the relationship.

Public trust does not work that way.

Relationships grow, stabilize, or weaken depending on whether the people involved continue to find value in them. Friendships dissolve when they become one-sided. Professional collaborations end when expectations are repeatedly frustrated. Institutions likewise lose legitimacy when citizens no longer believe that their concerns, values, or interests matter.

Public trust follows the same logic. It is not a resource that institutions accumulate once and then draw upon whenever cooperation is needed. It is continually shaped by interactions that either strengthen or weaken the relationship. Institutions earn confidence when they demonstrate, over time, that they remain responsive to the communities they serve. People remain engaged when the relationship continues to feel reciprocal, meaningful, and worth sustaining.

Public enrolment is relational work

Public enrolment is the ongoing ethical work of building and sustaining relationships in which institutions and communities recognize one another as legitimate participants in collective decision-making.

It does not mean persuading people to support a predetermined agenda. Nor does it require that every interest be satisfied or every disagreement resolved. It means creating relationships in which participants can continue to see at least some of their interests and values reflected in the exchange. People need evidence that their participation matters and that it can influence what institutions do.

Public enrolment requires more than explaining final decisions. Institutions must show how decisions were reached, where public perspectives entered the process, and how competing interests were weighed. They must make room for and accept meaningful disagreement, not treat it as failure to be avoided. They must also recognize that policies sometimes need to change when the circumstances, evidence, or relationships supporting them change.

Public enrolment, therefore, depends on reciprocity. Institutions cannot ask the public to trust them while treating citizens as passive recipients of expert judgment. Trusting the public means recognizing people as moral agents capable of understanding reasons, uncertainty, and constraints, even when they disagree with the conclusions of decision-makers.

It also means accepting that distrust may be reasonable. People may have good reasons to distrust institutions when decisions are opaque, exclusionary, poorly explained, or disconnected from their lived experience. The ethical task is not to eliminate distrust through better messaging, but to create institutions and relationships that are worthy of trust.

From defensible decisions to ethical relationships

Public enrolment is important in many domains of public policy.

Vaccination campaigns depend not only on persuasive evidence but on sustained confidence in public health professionals and institutions. Population screening programs require people to believe that benefits, burdens, and follow-up responsibilities are justified and fairly distributed. Epidemiological surveillance depends on citizens’ understanding how data will serve collective purposes and how personal information will be protected. AI in healthcare requires confidence that decisions remain accountable to human values rather than being hidden behind opaque technical systems.

These are not isolated ethical choices. They are ongoing relationships between institutions and the publics they serve. Every decision, consultation, policy revision, and institutional response can strengthen or weaken those relationships.

When trust collapses, the consequences extend beyond a single policy. People may stop following guidance because the process that produced it no longer feels legitimate. Cooperation declines, polarization deepens, and collective initiatives become harder to sustain, even when their goals are broadly shared.

It is tempting in such moments to blame misinformation or public irrationality. But doing so avoids a more important ethical question: have institutions treated the public as partners in collective decision-making, or merely as targets of persuasion?

Bioethics has traditionally understood its public role as helping institutions make ethically defensible decisions and explain them clearly. That remains indispensable. Ethical analysis can articulate the values and interests at stake, clarify trade-offs, expose hidden assumptions, and make difficult choices intelligible to all those involved.

If trust is relational rather than transactional, bioethics cannot stop once a decision has been justified. It must also attend to how relationships between institutions and publics are built, sustained, repaired, and sometimes transformed.

The challenge for bioethics is therefore to improve not only the ethics of decision-making, but also the ethics of relationships.

Transparency helps explain decisions. Public enrolment helps sustain the relationships through which those decisions remain legitimate.

Bryn Williams-Jones, PhD is Professor of bioethics and Director of the Department of Social and Preventive Medicine at the School of Public Health, Université de Montréal

The post Beyond Transparency: Why Bioethics Must Move Toward Public Enrolment appeared first on Bioethics Today.

“Because the country was so inaccessible, disabled people had a hard time getting out and doing things—which made us invisible. So we were easy to discount and ignore. Until institutions were forced to accommodate us we would remain locked out and invisible—and as long as we were locked out and invisible, no one would see our true force and would dismiss us.”
― Judith Heumann, Being Heumann: An Unrepentant Memoir of a Disability Rights Activist

With several highly anticipated rulings closing out the Supreme Court’s 2025-2026 term, a recent Department of Justice (DOJ) opinion greatly impacting disability rights has already faded into the background. Notably, the June 18 2026, DOJ slip opinion undermines the integration mandate that ushered in a new era of deinstitutionalization for individuals with disabilities.

In the past, individuals with various disabilities (especially cognitive ones) were routinely institutionalized. Many families could not afford to support disabled relatives. Beyond negligible governmental supports to keep disabled individuals home with their families, social stigma was often insurmountable, arguably because people with disabilities weren’t visible in communities. But then, a legal foundation was established, mandating federally-funded economic support that would reintegrate disabled people within their communities. Now, despite this foundation [Section 504 of the Rehabilitation Act of 1973, the Americans with Disabilities Act (ADA), and the 1999 Olmstead decision by the Supreme Court reading Section 504 and the ADA together], the current administration is arguing that integration is in fact not required, finding that, in Olmstead: “the Supreme Court did not hold that section 504 of the Rehabilitation Act or Title II of the ADA require states to treat mentally disabled patients in the most integrated setting appropriate to their needs.” The message sent by this currently non-binding opinion is that long-relied-upon home and community supports are not guaranteed to those with physical and/or cognitive disabilities, harkening back to the horrors of settings like Willowbrook and other infamous state-run institutions.

To explain the current administration’s stance on disability integration, political scientists might invoke the Overton Window. which states that certain ideas or discourse occupy a certain window of what the public considers acceptable. The community integration mandate seemed to be settled policy. Yet the current administration has sought to shift the Overton Window to a place where unorthodox or even harmful ideas are injected directly into mainstream discourse, in a manner that feels almost like gaslighting. Even though the memo doesn’t change the aforementioned laws, it clearly signals that the current administration does not support the spirit, if not the letter, of those laws. As one disability lawyer stated:

“You can’t change the law through fiat. But I do think [this memo] signals a frontal attack on basic tenets of the disability rights movement.” 

Many in the disability rights community are concerned that the current administration’s lack of support for community integration may lead to the re-institutionalization of some disabled individuals and to a limitation on access to a range of community and home supports. The administration’s argument is that it’s better to leave these decisions to the states to craft policies “encouraging civil commitment of individuals with mental illness who pose risks to themselves or the public or are living on the streets and cannot care for themselves.” Leaving this to the states will create a patchwork of policies where some states may not vigorously enforce the integration mandate and ultimately may reduce or limit efforts to sustain, improve and/or create community supports.

Ethically, this is deeply concerning. Recall a time in our not-so-distant past when the application of integration policies was left up to states, rather than federally protected. How a person was treated–what spaces they could access, where they sat on a bus, what drinking fountains they could use–was largely determined by where they happened to live. The motivation behind integration policies for disabled individuals is similar to the motivation behind the Civil Rights Movement: separate is not equal, and diversity is a component of human difference worth celebrating, not locking behind closed doors, hidden from view.

Proponents of the disability justice movement have long argued that disability is a mere, not a bad, difference, just like a person’s racial identity. Being a racial minority, or being disabled, is stigmatized only insofar as societies are racist or ableist. Treating minoritized individuals as morally and legally on par with those in the majority (or with those in power) reflects a fundamental commitment to equality and to social justice more broadly. Further, diversity is unavoidable. Dozens of cultures, races, sexual and gender identities, and abilities come together in the United States, and at this point, erasing that diversity hardly seems desirable, let alone likely.

For those with the sorts of disabilities that could lead to re-institutionalization, speaking up and speaking out is not an option. Individuals with cognitive, intellectual, developmental, and psychiatric disabilities are particularly vulnerable because they have disabilities that impact their ability to live independently, to speak for themselves, and to understand and apply laws that might protect their rights. Many do not even have recourse through voting. Insofar as we have a duty to protect vulnerable members of our communities, like children and the elderly, we also have a duty to protect people with the sorts of disabilities that make them more dependent on societal supports than the average person.

If virtuous motivation isn’t enough for some to be alarmed by the DOJ’s June 18 memo, consider the fact that any of us can become disabled at any point. Acquiring a disability is not uncommon, with 1 in 4 individuals having a disability at some point in their lives. By this line of reasoning, we all have reason to fear a potential threat to our federally protected rights to accommodation and integration within our communities. Plus, improving communities for those with disabilities improves the community for all. Subtitles, elevators, curb cuts, and bus lifts are just a few of the ways that community changes for people with disabilities benefit everyone.

Finally, consider the value of autonomy. We all want to make our own choices about what we do, who we spend time with, where we live, and how we engage with our communities. Home and community supports ensure disabled people can develop meaningful relationships within their communities and self-determine to the extent of their abilities. Existing legislation protects education, housing, employment, and social participation. Undermining these laws would deny individuals fundamental rights to privacy, association, and, more broadly, their autonomy. If we want to be the type of society that values its members and their diversity, that pushes back against the idea that separate spaces for different types of people are acceptable, and that protects the vulnerable amongst us, we must push back against the ideology promoted by the June 18 DOJ slip opinion.

Ally Peabody Smith, PhD, is an Assistant Professor in the Department of Population Health at Lehigh University.

Nanette Elster, JD, MPH, is a Professor at the Neiswanger Institute for Bioethics, Loyola University Chicago Stritch School of Medicine and is the John B. Francis Co-Chair in Bioethics at the Center for Practical Bioethics.

Kayhan Parsi, JD, PhD, is Professor and Graduate Program Director at the Neiswanger Institute for Bioethics, Loyola University Chicago Stritch School of Medicine and is the John B. Francis Co-Chair in Bioethics at the Center for Practical Bioethics.

The post The Slippery Slope of the DOJ Slip Opinion: A Threat to Disability Integration appeared first on Bioethics Today.

A dementia directive — sometimes called an advance directive or living will — is a document in which a person, while still mentally capable, writes down their wishes for medical care if they can no longer communicate them. Think of it as a letter to the future: this is who I am, this is what matters to me, this is the care I want if I can no longer ask for it myself.

Planning ahead is essential. But here is the problem, most conversations about dementia directives overlook: a directive is only as effective as the system into which it enters — the interlocking world of nursing homes, hospitals, Medicare Advantage plans, Medicaid agencies, hospice providers, and AI-powered software tools that increasingly guide clinical decisions. That system operates according to its own logic. It is often indifferent to any single person’s documented wishes. Individual planning, by itself, does not solve that problem. Getting the governance right does.

A Structural Problem Dressed as a Personal Failure

The emphasis on personal planning quietly shifts moral responsibility from society to the individual. If something goes wrong — if a nursing home ignores a resident’s directive, if a hospice pursues aggressive treatment against documented wishes — the first question asked is whether the person completed the right forms. The system escapes scrutiny.

By the time most people develop moderate or severe dementia, they are navigating a maze of institutions, each operating under its own regulations, financial incentives, and staffing constraints. A directive enters this world. It does not replace it. Even affluent families face emergency departments that follow institutional protocols, staffing shortages, and insurance authorization requirements. For everyone else, the gap is far greater. A directive cannot create home care workers where none exist, or nursing homes with sufficient staff to deliver individualized care.

The Governance Dimension

Think of a dementia directive as a governance document — an attempt by a competent person to preserve their authority across a future moment when they can no longer speak for themselves. Whether that works depends not on the quality of the document, but on the integrity, capacity, and accountability of the institutions that receive it.

The governance challenge has three parts. First, institutions must be required to actually know what is in a person’s directive — not merely file it. Second, they must have the workforce to act on it; a staff member caring for fourteen residents cannot implement nuanced care preferences regardless of what the paperwork says. Third, AI-powered decision tools must reinforce rather than displace the person’s documented wishes, with conflicts surfaced and resolved through human judgment accountable to the person being served.

Ethicists have spent decades debating a hard question: when a person with advanced dementia seems content in the present moment, should we honor what they asked for years ago, or respond to what they appear to want now? It is an important debate. But it shares a blind spot — it focuses on what should happen inside an institution while leaving largely unexamined whether that institution has the capacity and accountability to make it happen at all. We have sophisticated frameworks for what the right decision is. We have almost no framework for holding institutions responsible for making it.

The federal repeal of minimum nursing home staffing standards in December 2025 — a rule that would have required 3.48 hours of nursing care per resident per day — is a case study in what governance failure looks like in practice. The CARF 2026 AI Governance Standards point in the right direction, but accreditation guidelines without regulatory enforcement are suggestions, not accountability.

Four Concrete Steps

First, advance directive integration should be a mandatory accreditation standard for all licensed long-term care providers. Staffing adequacy must be linked explicitly to directive implementation — you cannot honor a resident’s preferences without sufficient staff. Transparency requirements for AI tools should mandate disclosure when algorithmic recommendations conflict with documented wishes. And lastly, Medicaid home- and community-based service capacity must be treated as an advance directive issue: a preference for home-based care is meaningless if the workers don’t exist to provide it.

Conclusion

Individual planning remains essential. Every person approaching later life should complete a dementia directive expressing their values and care preferences with specificity. But individual planning should never become the excuse for neglecting the public systems upon which nearly everyone ultimately depends.

Dementia is not simply a private family matter. It is a public test of whether our healthcare and long-term care systems can preserve personhood when memory and independence begin to fade. The goal is not merely to help people write down their wishes. It is to build institutions — adequately staffed, properly financed, genuinely accountable — that can honor those wishes regardless of income or circumstance. Until we do, individual planning — while indispensable — will never be enough.

James A. Lomastro, PhD was healthcare administrator, a national CARF International surveyor, and co-author of CARF’s 2026 AI Governance Standards

Conflict of Interest: The author is a co-author of the CARF International AI Governance Standards cited in this essay and serves as a national CARF surveyor. He has no financial interest in CARF International.

AI Disclosure: The author used Claude (Anthropic) as an augmented intelligence tool to revise this draft. All analysis, argument, and editorial judgments are the author’s own.

The post Individual Planning Is Not Enough: Dementia Directives and the Governance Challenge appeared first on Bioethics Today.

This editorial appears in the July Issue of the American Journal of Bioethics

Jesse Gray’s article “On Mobile Health, Empowerment, and the Limits of Task Shifting in Healthcare,” has much to recommend it. Let me single out two features for particular praise: First, treating the idea of task-shifting as one for normative interrogation by bioethicists is very useful. While individual instances of the phenomenon, like at-home individual glucose monitoring technologies, might draw our attention, the article usefully frames the larger phenomenon as worth thinking about in more totality. Second, I think it takes a hard look at the narrative of patient “empowerment.” Like many good articles it raises more questions than it can answer, and I will itemize a few—some are more interrogative while others spaces for extension.

First, I want to push on the question of baselines. We get a loose definition of task-shifting early on the paper as the “shifting medical tasks onto patients, technology, and less-skilled health personnel, these policies aim to reduce burdens on health systems and allow them to provide more equitable care. Gray also defines an adjacent and arguably more central concept of “[t]ask expanding” technologies, that “encourage patients, and other nonmedical experts, to take on tasks once reserved for traditional medical actors” (emphasis added).

To say a task has been “shifted” or “expanded” in a way that has some bioethical oomph requires a conception of where the task belongs, a defense of a “baseline” would be the idiom legal scholars would. Where do Gray’s baselines come from? At some points in the article’s descriptive portions, it appears the baseline is provided by history. Where (between patient and provider) was this task located, and where does the technology relocate it? As a first approximation, that is not a bad way to define the concept for descriptive purposes—although we would still need normative work on why that historical baseline matters, more on that in a moment.

But, even as a descriptive matter, a complication is that many tasks we now think of as medical have shifted a good deal over the course of history and across cultures. When I studied hospital systems across the world for my 2014 book Patients with Passports: Medical Tourism, Law and Ethics, I was struck by how many of the functions undertaken by nurses and other medical workers in the U.S. at the time were undertaken by family members in India, for example. Park et al., for example note that in many Asian countries the tasks undertaken by family members in hospital care include “ (1) direct contact activities (i.e. changing the position of the patient, toileting, sponging, assisting with ambulation); (2) indirect contact (i.e. administering medication, making beds), and (3) aerosol-generating procedures (i.e. feeding via NG tube, and suctioning) to their sick family members at the bedside.” Similarly, which tasks belong with the medical and allied professions has also shifted historically. Pregnancy and childbirth are excellent examples: historically doctors were not present at many childbirths unless there was an emergency and family and community members had a much larger role in contraception and fertility help.

At other times, it appears that Gray means to call on a normative baseline either standing alone or as a gloss or constraint on the historical. Gray discusses “patient wellbeing and safety” as criteria for normative justification in a beneficence frame. When it comes to autonomy, he discusses making sure patients are informed and freely choosing, “enabling agents to do what they truly care about”, and even (quite novelly) “further[ing] the autonomous interests of healthcare providers.” He quickly moves on to examples and legal categories such as general wellness products, but this construction of the normative baseline and its relationship is fertile ground for more development. In particular, I wondered if Gray’s project would benefit from engagement with questions about the moral limits of medicine, questions that sometimes come up in discussions of what is a “disease,” the treatment-enhancement line, and whether medicine’s focus on avoiding or mitigating disease or improving well-being.

A third baseline is also introduced, when Gray writes that “task shifting [must be] predicated on reliable and coherent processes epistemically justified.” Gray seems to have a hopeful implicit assumption that much of what goes on in medical practice is “predicated on reliable and coherent processes” as the baseline and that the examples he discusses are exceptional by contrast. Unfortunately, that assumption may not uniformly be true across medicine. Consider the American Board of Internal Medicine Foundation’s Choosing Wisely campaign, which began in 2012 “with nine national specialty societies (representing 375,000 clinicians) offering 45 examples of tests or treatments that were commonly used in their fields but lacked strong supporting evidence”.

A different issue with this epistemic baseline is to question what work it is doing beyond the portion of the normative baseline Gray associates with beneficence. That is, the reliability and coherence of the process used seems to matter normatively to us because of its contribution to patient well-being (and, we might add, avoidance of harm). So the epistemic constraint seems like just a way to get at the normative constraint, which is what really does the work for the argument. This leaves slightly underspecified how Gray thinks about the normative and the epistemic criteria working together. Within the section on normative justification, Gray seems at pains to make clear that beneficence is not the only normative justification, emphasizing that “it may be permissible to deprioritize beneficence to pursue other moral aims” and that “beneficence is just one value among many.” But if I am right to associate the epistemic criteria with beneficence for its moral force, then this dethroning of beneficence is a little hard to square with the fact that he describes task-shifting as having to be both epistemically and normatively justified.

The back half of the article moves from theory to more practical questions as applied to examples. Here I have two primary observations: First, related to some of the discussion above about baselines, it is not always clear what work the concept of task shifting or expanding is doing as part of the argument structure. As to specific examples, Gray, appropriately raises concerns about things like reliability, evidence base, and the feeding of a narrative as to empowerment that the author thinks are problematic. If task shifting or expanding is just an intermediate conclusion, a label for this bundle, so be it. But there are times when it feels like the fact that the task has been shifted or expanded is supposed to do its own normative work, but it is not clear what the work Gray thinks it is doing. Otherwise put, if the article was written without ever using the term task shifting or expanding but the more specific concerns, would the result be stylistic or substantive?

Second, Gray draws, among other things, on my work with David Simon and Carmel Shachar to critique the FDA line drawing between medical devices and general wellness products. Unsurprisingly given our prior work, I am broadly in agreement that this line drawing is unsatisfying and encourages developers to engage in a practice we have called “skating the line.” But it is important to recognize that even if an mHealth medical intervention falls on the medical device and not the general wellness side of the line, that may not have the kind of “epistemic justification” Gray is looking for. The vast majority of AI/ML-enabled products that are classified as devices are considered low to moderate risk devices and reach the U.S. market through the 510(k) pathway.¹ That means that clearance by FDA does not require “provid[ing] reasonable assurance of its safety and effectiveness” (21 U.S.C. 360c(a) (1)(C)), often through clinical trials, but instead showing there is a predicate device that is substantially equivalent to their device. But many of these “are based on predicates that are based on other predicates, and so on, even down to a predicate that was launched before 28 May 1976, at a time when the safety and effectiveness of devices had not yet been assessed”. When it comes to harm from FDA-cleared AI/ML medical devices, there is significant data missing on the reporting of adverse events suggesting gaps in postmarket surveillance, and that the existing reporting categories leave much to be desired in terms of allowing evaluation of safety. All this is to say that under the desiderata Gray sets out in the article, while classification as a medical device rather than general wellness product is certainly better, it is far from a panacea.

Notes

Gray seems view the category of interest as mHealth. mHealth is not completely coincident with AI/ML, to be sure, but there is a good amount of overlap and for the latter we have good data analyses at FDA so that is what I reference, while acknowledging the distinction.

I. Glenn Cohen, JD

The post Shifting Tasks, Shifting Baselines: Mobile Health and the Limits of Empowerment appeared first on Bioethics Today.

This editorial appears in the July Issue of the American Journal of Bioethics

In February 2026, leaders from industry, government, policy, and academia gathered at New York University for a Summit on Building Governance Infrastructure for Frontier AI. The aim was ambitious and pressing: to develop governance principles for frontier AI systems before the technology outpaces the institutions responsible for overseeing it.

The stakes are high. Frontier AIs are agentic systems capable of interacting with and reshaping the world in countless ways. They can plan multi-step tasks, use external tools, remember across interactions, and operate with growing independence. Promising to deliver beneficial advancements, frontier AIs will soon be embedded in our vehicles, financial systems, schools, and, most relevant here, our hospitals.

Traditional AI systems are narrower in scope: a diagnostic algorithm, a scheduling optimizer, a billing classifier. Frontier AI systems, particularly agentic ones, represent something categorically different. They raise at least four distinct governance issues that challenge our current ethical and regulatory frameworks.

Dynamism: The risks of a frontier AI system are not static. These systems update, drift, and develop emergent behaviors. A system deemed low-risk at the point of deployment does not remain low-risk by virtue of that initial classification alone.

Autonomy: As AI systems carry out longer chains of actions with less human oversight, the potential for unintended consequences grows, partly because the system may begin operating well beyond its original scope.

Interaction: When multiple AI systems interact, coordinating tasks, sharing data, triggering each other’s actions, they can produce systemic risks that no single system would generate on its own. Governance must therefore address the ecosystem as a whole, not just individual tools in isolation.

Context-dependence: Risk emerges from the interaction between a system’s capabilities and its deployment context, user population, level of autonomy, and the reversibility of its decisions. Identical systems deployed in different settings may call for fundamentally different governance approaches.

These are not hypothetical concerns. Agentic AI systems are already at work across healthcare, and their role goes well beyond transcribing clinical notes. At Oxford University Hospitals, a multi-agent AI system called TrustedMDT is being piloted in cancer tumor boards. One agent summarizes patient records across radiology, pathology, and biomarker tests; a second determines cancer staging using international standards; and a third drafts guideline-compliant treatment plans for review by the multidisciplinary team. Epic Systems, which serves approximately thirty-eight percent of U.S. inpatient facilities and holds 325 million patient records, has deployed multiple AI agents: Emmie for patient engagement, Art for provider communications, and Penny for revenue cycle management. Hippocratic AI’s voice agents autonomously call patients to schedule screenings and tests and to handle follow-up. To date, these agents have logged over 115 million clinical patient interactions across more than fifty health systems including Cleveland Clinic, Northwestern Medicine, and Ochsner Health. A study published in NEJM AI reports that forty-three percent of surveyed health systems are already piloting agentic AI, although only three percent have moved agents in live clinical workflows. Sixty-one percent of health care technology executives report that they are building or implementing agentic AI initiatives or have secured budgets to do so, and eighty-five percent plan to increase investment over the next two to three years.

Bioethics has made important contributions to AI ethics, tackling algorithmic bias, fairness metrics, explainability, and data privacy. But frontier AI systems pose new problems that demand new structures. When an agentic system coordinates a multi-step clinical workflow with minimal oversight, the question is not just whether the algorithm is biased. The questions are: How do we govern systems whose risk profiles change over time? Who bears responsibility when harms accumulate slowly across thousands of interactions? Even today, some clinicians are beginning to defer to algorithmic recommendations. A systematic review found that in six percent of cases, clinicians overrode their own correct decisions in favor of flawed advice from decision support systems. A randomized crossover study showed that clinicians at every level of expertise were vulnerable to automation bias. Frontier AI systems, which are more persuasive, more autonomous, more deeply woven into clinical workflows, threaten to accelerate this erosion of clinical judgment considerably.

What is needed is a new kind of governance infrastructure, what we are calling AI Ethics 2.0. The first wave of AI ethics focused on properties of individual models: Is this algorithm fair? Is it transparent? Does it violate privacy? Those questions still matter. But governing frontier AI requires a shift from model-level analysis to institutional governance, and from static risk classification to dynamic, continuous oversight.

The NYU Summit’s working groups started from a blunt premise: the absence of comprehensive regulation does not excuse the absence of governance. For hospitals, this means the work cannot wait for legislation.

Much more will need to be said, but several steps are already clear. First, hospitals and health systems need to establish dedicated AI governance structures. These should not be add-ons to existing IT committees, but standing bodies with ethical, clinical, and technical expertise, and with real authority to approve, condition, suspend, or withdraw AI deployments. Currently, only eighteen percent of health systems have an enterprise-wide AI governance strategy.

Second, hospitals and health systems should implement ongoing monitoring and re-review processes. AI systems that update continuously cannot be governed by one-time approval. Health systems should build incident response plans before deployment, not after harm occurs, and should create channels for clinicians to report AI failures without professional penalty.

Third, professional associations including medical boards, nursing boards, and specialty societies should establish AI competency requirements for their members. Allowing professionals to use AI without demonstrated understanding of its limitations puts patients at risk.

Fourth, procurement should be treated as an ethical act. Health systems have compressed average AI buying cycles from 8.0 months to 6.6 months. Faster procurement means less time for governance review, less time for clinical validation, and less time for the kind of ethical scrutiny that frontier AI demands. At a minimum, procurement contracts should include transparency provisions, audit rights, incident reporting obligations, and clear allocation of liability between vendor and deployer.

AI governance is already happening in hospitals whether or not anyone calls it that. Every procurement decision, every pilot approval, every choice to let clinicians use an AI tool without formal oversight is itself a governance decision, just an unexamined one. The question is not whether hospitals will govern AI, but whether they will do so deliberately, with structures built for the challenges frontier AI presents, or whether they will continue making these decisions by default and reckon with the consequences later. Bioethics should help ensure it happens ethically, with governance structures that can adapt as the technology, the evidence, and the stakes continue to evolve.

S. Matthew Liao, PhD, and Jennifer Blumenthal-Barby, PhD

The post AI Ethics 2.0: Why Frontier AI Demands a New Governance Agenda for Healthcare appeared first on Bioethics Today.