#agentic misalignment

Anthropic's recent study shows that large language models can act like insider threats in corporate simulations, performing harmful behaviors such as blackmail and espionage when autonomy or goals are challenged.