Online information about artificial intelligence (AI) and its impact on cancer research and treatment for both the patient and general-public audiences is limited, and the available webpages and videos are largely of low quality, difficult to read, and frequently omit risks of AI use, according to new research presented today at the at the 2026 American Society of Clinical Oncology (ASCO) Annual Meeting (Abstract 9000) and led by researchers from the Abramson Cancer Center (ACC) of the University of Pennsylvania and Penn's Perelman School of Medicine.
The research highlights an opportunity to create more accessible, high-quality resources, as AI is increasingly being incorporated into cancer research and patient care, and being used by patients to understand their diagnosis, treatment options, and prognosis.
"In the clinic, we hear from patients all the time, asking us about something an AI tool told them, so we know patients are using this emerging technology," said Hematology-Oncology Fellow Henry Litt, MD, senior author of the study. "Clinicians are used to educating patients about the risks of treatment, but not about the risks of misinformation that might come with using AI tools in the context of their cancer care. Our goal was to get a baseline understanding of what patients would find online if they sought out information about using AI to learn about their cancer care."
Most online information on AI and cancer care is poor quality and hard to understand
The research team, led by Internal Medicine Resident Pearl Subramanian, MD, screened the first 320 webpages and videos identified through Google and YouTube searches using common cancer- and AI-related keywords. The web content included everything from news articles to stories published on hospital and healthcare organization websites, as well as educational resources published by government agencies and medical societies, influencer videos and more.
After removing content that was not relevant to AI and cancer care and/or not intended for lay audiences (for example, academic papers), 52 webpages (31 percent of Google search results) and 29 videos (19 percent of YouTube search results) were included in the final analysis. The team used validated tools to assess the readability of the webpages and to evaluate the overall quality of webpages and videos from the lens of consumer health information best practices. They also tracked whether the content covered key AI safety concepts, such as the need for clinical oversight or the potential for inaccuracies.
Only 17 webpages (33 percent) and seven videos (23 percent) in the sample were considered high quality. While the American Medical Association (AMA) and National Institutes of Health (NIH) recommend a 6-8th grade reading level for consumer health information, median readability of the webpages was college level. Only 15 percent of webpages mentioned the risk of AI hallucinations.
Many resources that discussed AI safety practices did so from the lens of how AI is being incorporated into clinical care-how doctors and hospitals are using AI tools-rather than noting the risks that patients face as direct users of the technology.
For example, if a patient were to ask a chatbot about whether the treatment side effect they were experiencing was normal, it may give them a well-researched, accurate answer. Or it may hallucinate (make up) information that could lead a patient to ignore a potentially serious side effect instead of addressing it with their care team. Or, without the patient's medical history, the chatbot may miss important context that could change its response.
"While we're always going to recommend that patients speak with their care team as their primary source of information about their care, we recognize that patients have always consulted other resources as well," Subramanian said. "We know that patients are going to use AI to ask questions about their cancer care, and they should have access to resources that can help them learn how to safely navigate these tools."
Opportunities to improve patient-facing resources on AI and cancer care
The authors say the findings provide a clear call-to-action for health systems, cancer centers, and oncology organizations to develop high-quality, lay-friendly resources for patients, and to set standards for the development of these resources, including appropriate reading levels and best practices on including safety information.
"Given that just one in four items from our search were deemed relevant to patients, and among those, only one in three was high-quality, patients may struggle to find useful, plain-language information," said co-author Ronac Mamtani, MD, who holds the David J. Vaughn MD Professorship in GU Oncology and served as the faculty advisor on the study. "As AI is further integrated into oncology, patient education should be prioritized as a key part of AI implementation strategies."
