LodeHQ
AI news for people who shipFree, by email

AI · arxiv.org · Monday, 16 November 2026

Loss framing pushed an LLM toward collusion, but hidden states reversed it

In 260 simulated London silver-fixing negotiations, loss framing raised Gemma 3 12B's collusion from 5% under neutral framing to 25%, while blame raised it to 33%. The researchers then injected text-induced internal states at layer 22: loss and blame states made the agent refuse all 60 requests, but gain and credit states lifted collusion to 32% from 13% with no injected state. Management pressure increased trading volume 10 to 19 times without increasing collusion.

Read the source: arxiv.org ↗In the 16 November issue →