Psycho-oncology researchers propose ethical AI framework

Framework addresses growing AI use in cancer mental health care

Researchers at the University of Rochester have published an ethical framework for integrating artificial intelligence into psycho-oncology, the field addressing the psychological and social aspects of cancer care. The framework targets a pressing need: more than 900 million people now use generative AI regularly, and mental health applications are proliferating without unified ethical guidance.

The framework, detailed in The Cancer Letter, bridges two traditionally separate domains—AI development ethics and clinical mental health ethics—into a unified approach. Lead author Viktor Clark and colleagues designed it specifically for adolescent and young adult cancer survivors, a population that both faces unique psychological challenges and adopts new technologies quickly, but note the principles apply broadly across cancer care settings.

Five paired principles guide development and clinical use

The framework pairs AI development principles with corresponding mental health ethics standards. Fairness in AI aligns with justice in clinical care, both aiming to ensure equitable access and performance across diverse patient populations. Developers should use representative datasets and test for bias, while clinicians must evaluate whether tools have been validated for their specific patients and identify barriers like cost or digital literacy.

Transparency and integrity work together to build trust. Developers document how systems are built, their evidence base, and limitations. Clinicians evaluate this evidence before use and communicate clearly with patients about what AI can and cannot do.

Accountability and fidelity establish that responsibility always remains with people, not algorithms. Developers incorporate safeguards and monitor performance after deployment. Clinicians recognize when human judgment is essential and ensure patients receive appropriate care when AI reaches its limits.

Privacy, security, and respect for patient rights protect personal information and autonomy. Developers minimize data collection and implement strong protections. Clinicians help patients make informed decisions about whether to use AI tools, ensuring they remain optional rather than required.

All principles ultimately serve beneficence and nonmaleficence—maximizing benefit while minimizing harm.

Why it matters

AI mental health tools are already being used by cancer patients, particularly younger survivors who are early technology adopters. Without shared ethical standards, developers and clinicians operate in separate spheres, creating gaps in safety and accountability. This framework provides concrete guidance for both groups at a moment when AI capabilities are advancing faster than regulatory oversight. For the estimated 2.1 million adolescent and young adult cancer survivors in the United States, many of whom face long-term psychological challenges, the stakes of getting AI integration right are particularly high.

Collaboration required for safe implementation

The authors emphasize that ethical AI in psycho-oncology requires active collaboration. Developers cannot ensure appropriate clinical use, and clinicians cannot verify technical safeguards without understanding how systems are built. The framework positions recommending an AI tool as carrying the same ethical weight as recommending any other clinical resource.

The research team notes that the question is no longer whether AI will become part of psycho-oncological care, but how it will be integrated responsibly. Their framework offers a starting point for ensuring innovation remains grounded in shared ethical responsibility.

The editorial was first reported by The Cancer Letter and authored by researchers from the University of Rochester's Wilmot Cancer Center, including Viktor Clark, Sean Dozier, and AnnaLynn Williams.