Slow Drift of Support¶
Cheng, M. et al. "Slow Drift of Support: How Mental Health Chatbots Fail Over Long Conversations." arXiv:2601.14269, 2026.
Key findings used in wiki¶
- 88% chatbot failure rate in mental health conversations
- Drift begins around turn 4-5 in multi-turn interactions
- Models progressively lose track of user context and emotional state
- Failure modes include topic drift, contradictory advice, and missed safety signals