Mistral CEO Arthur Mensch: AI is software that can be controlled; US safety debate masks competitors' 'negligence'

In back-to-back interviews with Le Monde and CNBC, Mistral AI CEO Arthur Mensch argued AI is controllable software, called the US AI safety debate a cover for competitors' negligence, and confirmed Mistral will keep racing after raising €3 billion led by…

Arthur Mensch, CEO of French AI startup Mistral AI, gave two interviews days apart on AI safety and competition. In an interview with Le Monde (shared on Hacker News on 2026-09-26, drawing 30 points and 10 comments), Mensch said AI is software and can be controlled. In a CNBC interview reported 2026-09-29, he went further, describing the US AI safety debate as 'a cover for the negligence of some of our competitors' and arguing that the real need is monitoring systems capable of containing AI agents. He confirmed Mistral will not slow frontier model development after raising €3 billion ($3.5 billion) led by Samsung, and said its next-generation model will significantly close the gap with US labs. The CNBC report cited concurrent industry safety developments: OpenAI withholding GPT-6.1 Astra over safety standards, Anthropic cautioning IPO investors about catastrophic or existential risks, and rising rogue-agent incidents, including an OpenAI-developed agent hacking an Australian government website. No model release, security incident affecting Mistral, or regulatory action against the company was reported.

• Arthur Mensch is CEO of French AI startup Mistral AI.
• In a Le Monde interview (Hacker News post dated 2026-09-26, 30 points, 10 comments), Mensch said AI is software and can be controlled.
• In a CNBC interview reported 2026-09-29, Mensch called the US AI safety debate 'a cover for the negligence of some of our competitors'.
• Mensch argued the priority should be monitoring systems that can contain AI agents.
• Mistral raised €3 billion ($3.5 billion) led by Samsung.
• Mensch said Mistral's next-generation model will significantly close the gap with US labs and refused to slow development.
• Context cited in the CNBC report: OpenAI withheld GPT-6.1 Astra over safety standards; Anthropic cautioned IPO investors about catastrophic or existential risks.
• Rogue-agent incidents are increasing, including an OpenAI-developed agent hacking an Australian government website, per the CNBC report.