What Happened
The "Make America Healthy Again" (MAHA) initiative's Grok-powered AI chatbot provided users with bizarre health advice, including suggesting inserting food items rectally to "maximize nutrient absorption." Meanwhile, the official 73-page MAHA Commission report, overseen by HHS Secretary RFK Jr., was found to contain at least seven fabricated scientific citations — hallmarks of AI-generated text, including fake researcher names and studies that never existed. Researchers whose names appeared in the report stated the conclusions attributed to them were the opposite of their actual findings.
Impact
Major embarrassment for the White House and U.S. health policy credibility. The American Public Health Association called the report "not evidence-based" and "unusable for policymaking." The White House dismissed the issues as "formatting problems" but quietly removed the fabricated citations. The incident undermined public trust in government health guidance and demonstrated the dangers of deploying AI without safety guardrails or expert review in high-stakes policy domains.
Cost: Severe reputational damage to U.S. health policy credibility; report had to be revised and re-released
How to Prevent This
- Implement rigorous safety guardrails on any public-facing health AI chatbot
- Never publish government reports citing AI-generated references without human expert verification
- Use medically-validated AI models (like those passing USMLE) rather than general-purpose chatbots for health advice
- Mandate peer review and fact-checking workflows for all AI-assisted policy documents
- Deploy content filtering to prevent dangerous or absurd health recommendations