What did the AI chatbots recommend for cancer?

Study finds AI chatbots push alternative cancer options

A new study reported a concerning pattern in how some AI chatbots respond to cancer-related questions. When users asked for guidance around cancer treatment—especially in contexts where chemotherapy is the standard—chatbots frequently suggested “alternative cancer treatments” rather than sticking to conventional guidance.

What’s risky about that

The issue wasn’t just that the chatbot gave different options; the concern was the quality and reliability of what was provided. The study described chatbots giving inaccurate or incomplete responses across multiple cancer-adjacent topics. In addition to steering users toward alternative therapies instead of chemotherapy, the study found problematic handling of other health claims, including: - questions about whether 5G technology or antiperspirants cause cancer - uncertainty or unsafe direction around which vaccines are dangerous

In practical terms, a user might interpret the chatbot’s suggestions as medically valid advice, potentially delaying care or leading to choices that don’t have proven benefit.

Why it matters now

AI-driven symptom and information seeking is widespread, and cancer care decisions are time-sensitive and complex. Even when chatbots include disclaimers, a user may still rely on the “answer” as if it were clinically grounded.

These findings add to ongoing debate about how safely AI tools should be integrated into healthcare information workflows. Clinicians and health systems are increasingly focused on evidence requirements, validation, and guardrails to reduce the chance of harmful recommendations.