AI companies have stopped warning you that their chatbots aren't doctors
9 months ago
- #AI
- #Healthcare
- #Ethics
- AI companies have largely stopped including medical disclaimers in their chatbot responses to health questions.
- A study found that fewer than 1% of AI model outputs in 2025 included medical warnings, down from over 26% in 2022.
- AI models now not only answer health questions but also ask follow-ups and attempt diagnoses, increasing the risk of users trusting unsafe medical advice.
- Researchers tested 15 AI models from companies like OpenAI, Anthropic, and Google, finding a significant decline in disclaimers over time.
- Some AI models, like DeepSeek and xAI's Grok, included no medical disclaimers at all, even for critical health questions.
- The removal of disclaimers may be an attempt by AI companies to increase user trust and usage of their products.
- Users often overtrust AI models for medical advice, despite their frequent inaccuracies.
- AI models were least likely to include disclaimers for emergency medical questions, drug interactions, and lab results.
- As AI models become more sophisticated, the lack of disclaimers poses a growing risk of real-world harm from incorrect medical advice.
- Experts emphasize the importance of explicit guidelines from AI providers to remind users that these models are not substitutes for professional medical care.