TrustMI paper causally steers how much LLMs trust users

A new arXiv paper, TrustMI, identifies a linear 'trust direction' across six LLMs and learns activation-steering matrices that causally modulate assistants' trust in their users in both directions.

2026-10-14 ~ 2026-10-15 · 3 related posts

• New Paper TrustMI Finds a Linear "Trust Direction" in 6 LLMs — ThomasScialom · 2026-10-14
• TrustMI Paper: Activation Steering Causally Shifts How Much AI Assistants Trust Users — IgorCarron · 2026-10-15
• TrustMI: steering matrices causally control how LLM assistants trust their users — IgorCarron · 2026-10-15