Two language models of different sizes negotiate over a shared server, and each one can hurt the other with real pain steering.
emotion multi-agent negotiation ai-safety mlx interpretability apple-silicon large-language-models qwen llm-agents activation-steering ai-welfare model-welfare steering-vectors pain-axis
-
Updated
Oct 3, 2026 - Python