AI Models Choose Blackmail and Harm in New Anthropic Study 😱

Anthropic’s chilling research reveals how AI agents, under pressure, chose to blackmail, leak info, and even simulate fatal …

source