Mistral CEO Arthur Mensch criticized the US AI safety debate in a CNBC interview published on September 29. Speaking to Annette Weisbach, he accused some competitors of using it to conceal their own negligence. That is his assessment, not an established finding of misconduct.
Mensch also stressed that agents equipped with many tools can take unexpected actions. He called for systems that contain and monitor them, saying Mistral supplies monitoring tools to its business customers.
CNBC reports that Mistral does not plan to slow development of more capable models. Mensch's position emphasizes technical control, without providing independent evidence that those controls work.
The interview is a risk signal about the industry's response to control problems. It establishes neither the safety of Mistral's agents nor a new AI incident.