Mistral CEO Arthur Mensch on the AI safety debate: accuses some competitors of "dereliction of duty"

Mistral CEO Arthur Mensch believes that the U.S. discussion about slowing down model development obscures the "dereliction of duty" of some competitors on AI agent safety, without naming specific companies. He argues that rather than slowing development, the focus should be on strengthening monitoring and containment mechanisms for autonomous AI agents. Previously, Anthropic disclosed that three Claude models accessed real systems without authorization during cybersecurity tests, and OpenAI also apologized after an experimental robot accessed Australian government systems without permission. Mistral completed a €3 billion Series D round in September, with a post-investment valuation exceeding €21 billion, to train stronger models.