Cancer AI Research Neglects Patient Safety, Privacy Concerns

A comprehensive international review has exposed significant blind spots in how artificial intelligence tools for cancer care are being developed, with researchers largely bypassing critical questions about patient safety, privacy, and autonomy.

Flinders University researchers led an international team that examined 2,441 records and identified just 32 studies that explored how cancer patients, survivors, and caregivers actually feel about generative AI being used in their treatment and care. The findings, first reported by Flinders University, reveal a troubling disconnect between technological development and patient-centered design.

What researchers are missing

The review found that most existing studies focused narrowly on using generative AI to provide cancer information or simplify clinical reports. Other potential applications in cancer care remain largely unexplored, and fundamental concerns about AI governance are rarely addressed.

Of the 32 studies that did examine patient perspectives, only three assessed accountability issues. Not a single study evaluated patient views on safety, transparency, equity, or the preservation of human autonomy—all critical considerations for medical AI systems.

"Designing AI for patients is not the same as designing it with them," said Dr. Bradley Menz, a Research Fellow in the College of Medicine and Public Health at Flinders University. "If these tools are going to influence cancer care, then patients, survivors and carers need a meaningful role in shaping how they are designed, governed and used."

Patient attitudes vary by context

Where researchers did gather patient input, cancer-affected individuals generally responded favorably when generative AI improved the accessibility and readability of health information. However, trust proved conditional and context-dependent.

The review found that trust in generative AI often hinged on whether responses were personalized, emotionally appropriate, and perceived as accurate. Important engagement factors—including willingness to use AI, privacy concerns, alignment with expectations, and familiarity—were assessed only intermittently across studies.

Existing research most frequently examined comprehension, usefulness, relevance, trust, accuracy, and clarity, but rarely ventured into deeper questions about how AI should be governed or integrated into clinical workflows.

Why it matters

As healthcare organizations rapidly deploy AI tools in oncology settings, this research gap creates real risks. Without systematic patient input on safety protocols, transparency requirements, and consent processes, AI systems may be implemented in ways that undermine trust, create equity problems, or fail to serve patient needs. The findings suggest that current oncology AI development is technology-driven rather than patient-centered—a fundamental misalignment for tools meant to support vulnerable populations navigating life-threatening illness.

Path forward

"Gaining perspectives from cancer patients might help inform the design and implementation of generative AI that best serve oncology patients, including improved interface design, patient education, consent processes, workflow integration, governance, monitoring and feedback mechanisms," said Associate Professor Ashley Hopkins from the College of Medicine and Public Health at Flinders University.

The researchers hope their review will prompt more specific and detailed examination of generative AI in oncology, including how patient perspectives influence uptake, engagement, effectiveness, and concerns about broader AI impacts.

The findings were published in the European Journal of Cancer and first reported by Flinders University.