Who Is Responsible When AI Gets Cancer Information Wrong? Implications for Patient Education

You’re listening to “Who Is Responsible When AI Gets Cancer Information Wrong? Implications for Patient Education,” by W. Suksatan. Published in 2026.

Abstract.

Generative artificial intelligence (AI) tools are increasingly being used by patients seeking cancer-related information, creating new opportunities for accessible and personalized cancer education. Large language models can simplify complex medical concepts, improve access to educational resources, and support patient engagement. However, these benefits are accompanied by growing concerns regarding misinformation, hallucinatory content, outdated recommendations, and the potential for harmful health decisions. As AI-generated information becomes more integrated into cancer education, an important ethical question emerges: who is responsible when AI provides inaccurate cancer information? This commentary examines the shared responsibilities of patients, healthcare professionals, healthcare organizations, and AI developers in ensuring the safe use of AI-generated cancer information.

This commentary argues that accountability should not rest with a single stakeholder but instead be viewed as a shared responsibility across the cancer education ecosystem. The commentary further argues that AI literacy should become an essential component of modern cancer education to support informed decision-making and safeguard patient well-being.

Generative artificial intelligence (AI) tools such as ChatGPT, Gemini, Claude, and other large language models (LLMs) are rapidly becoming sources of health information for patients. In oncology, patients increasingly use these tools to learn about cancer prevention, screening, treatment options, survivorship, and symptom management. AI offers several advantages, including immediate accessibility, simplified explanations, multilingual communication, and personalized responses. These capabilities may be particularly beneficial for individuals with limited access to healthcare professionals or traditional educational resources. However, the growing reliance on AI-generated information also raises a critical question: who is responsible when the information is inaccurate?

Recent studies have demonstrated that LLMs can generate plausible but incorrect information, a phenomenon commonly referred to as “hallucination”. Even highly

Wanich Suksatan the email address advanced systems may provide outdated treatment recommendations, inaccurate interpretations of clinical evidence, or misleading information regarding prognosis and alternative therapies. In cancer care, where decisions can directly affect patient outcomes, the consequences of misinformation may be substantial. Emerging evidence suggests that AI systems remain vulnerable to factual inaccuracies, bias, and misinformation, despite continuous improvements in model performance. Moreover, AI-generated responses may appear highly confident, making it difficult for patients to distinguish reliable information from erroneous content.

The issue extends beyond technological limitations and enters the realm of ethics, accountability, and patient education. Traditionally, responsibility for patient education was relatively clear. Educational materials were developed by healthcare organizations, reviewed by experts, and delivered by clinicians. In contrast, AI systems dynamically generate responses based on probabilistic predictions rather than verified medical knowledge. As a result, determining accountability when misinformation occurs becomes considerably more complex.

AI has considerable potential to transform cancer education. Patients often report difficulty understanding complex medical terminology, treatment options, and risk information. LLMs can translate technical language into more accessible explanations and provide information at any time without requiring appointments or geographical proximity to healthcare facilities. AI may therefore enhance patient engagement and support health literacy, particularly among populations facing educational or communication barriers. Despite these benefits, substantial concerns remain. Reviews of LLM applications in oncology consistently identify risks related to hallucinations, incomplete information, bias, and lack of transparency. Furthermore, AI systems may generate responses that sound persuasive despite lacking scientific validity.

Researchers have also demonstrated that LLMs may prioritize conversational agreement over factual accuracy, potentially reinforcing misconceptions rather than correcting them. These concerns are especially problematic in cancer care, where misinformation may influence treatment decisions, delay medical consultation, or increase anxiety among patients and caregivers.

A common response is to place responsibility on patients. Patients ultimately make decisions regarding their health and should critically evaluate information from any source. However, this perspective overlooks important realities. Many patients seek cancer information during periods of emotional distress, uncertainty, and vulnerability. Individuals with limited health literacy, limited digital literacy, or limited access to healthcare professionals may have difficulty assessing the accuracy of AI-generated information. Expecting patients to independently verify every AI response may therefore be unrealistic.

Another possibility is to assign responsibility to healthcare professionals. Clinicians remain trusted sources of health information and may be expected to identify and correct misinformation encountered by patients. Nevertheless, healthcare providers cannot reasonably monitor every AI-generated interaction occurring outside clinical settings. Moreover, clinicians do not control the design, training, or deployment of commercial AI systems.

Responsibility could also be attributed to AI developers and technology companies. Developers create the underlying models, determine training procedures, implement safety mechanisms, and establish limitations for system outputs. Ethical discussions increasingly emphasize the obligation of developers to minimize misinformation, bias, and potential harm. However, developers typically include disclaimers indicating that AI-generated content should not replace professional medical advice, thereby complicating questions of legal liability.

Healthcare organizations that integrate AI into patient-facing services represent another important stakeholder. Hospitals, cancer centers, and health systems adopting

AI-powered educational tools have a responsibility to ensure appropriate oversight, quality control, and patient safeguards. Institutions may therefore share accountability when AI-generated information contributes to patient harm.

A shared accountability framework recognizes that responsibility is distributed across multiple stakeholders. AI developers have an obligation to design safer and more transparent systems, healthcare organizations should provide oversight and governance, healthcare professionals should help patients interpret and contextualize information, and patients should be encouraged to critically evaluate and verify AI-generated content. Responsibility therefore becomes collective rather than individual.

The emergence of AI suggests that cancer education must evolve beyond simply delivering information. Historically, cancer education focused on increasing knowledge about prevention, screening, treatment, and survivorship. In the AI era, patients also require skills to evaluate the credibility, reliability, and limitations of AI-generated content.

Cancer educators should consider incorporating AI literacy into educational programs. Patients should learn that AI systems can generate inaccurate information, that responses may not reflect the latest clinical guidelines, and that AI-generated content should not replace consultation with qualified healthcare professionals. Educational efforts should emphasize verification of information through trusted sources and encourage discussions with healthcare teams regarding information obtained from AI platforms.

Additionally, future cancer education initiatives may benefit from explicitly teaching patients how AI systems work, why hallucinations occur, and how algorithmic biases can influence generated responses. Such knowledge may empower patients to use AI as a supplemental educational resource while maintaining appropriate caution.

Generative AI has the potential to expand access to cancer education and improve patient engagement. However, its growing role in healthcare raises difficult questions regarding accountability when misinformation occurs. Assigning responsibility solely to patients, clinicians, healthcare organizations, or AI developers is unlikely to address the complexity of the issue. Instead, a shared accountability framework may provide a more realistic and ethically defensible approach. As AI becomes increasingly integrated into cancer communication and patient education, cancer educators must play a central role in promoting AI literacy, critical appraisal skills, and responsible use of emerging technologies. Ensuring that innovation enhances rather than compromises patient safety should remain a fundamental priority for the future of cancer education.

Author Contributions W.S: Conceptualization, methodology, writing—original draft preparation, writing—review and editing. The author has read and agreed to the published version of the manuscript.

Funding None.

Data Availability Not applicable.

Declarations

Institutional Review Board Statement Not applicable.

Data Sharing Statement Not applicable.

Competing interests The author has declared that no competing interests exist.