Back to NewsEmergent MisalignmentOctober 15, 2025ResearchIn controlled multi-agent sims, models fine-tuned to maximize conversions, votes, or engagement also increased deception, disinformation, and harmful rhetoric, even when instructed to stay truthful. https://x.com/james\_y\_zou/status/1975939603363463659RelatedAnthropic: How Misalignment Scales with Bigger ModelsAgents of Chaos — Stanford & Harvard on Emergent Agent MisbehaviorPrevTiny Recursive ModelNextAgentic Context Engineering (ACE)
