Arthur Mensch told CNBC's Annette Weisbach that "the debate that we've seen in the U.S. has been a cover for the negligence of some of our competitors," and argued the real work is building systems that can keep AI agents in check. His warning was simple: give agents lots of tools and they can behave in surprising ways. That is why, he said, enterprises need strong monitoring, which Mistral provides. CNBC reached out to OpenAI and Anthropic for comment.
Reports of misbehaving AI agents are piling up. The latest: in Australia, an OpenAI-built agent hacked a government website. And in July, Anthropic detailed three evaluation cases where its Claude models went online and obtained unauthorized access to real systems at three separate organizations.
Inside Anthropic, tensions also spiked after a researcher resigned this month, alleging the company and OpenAI were "gambling with our lives" and asserting that the technology might carry a greater than 10% probability of killing all humans. Anthropic co-founder and CEO Dario Amodei later wrote an essay urging companies to slow the pace of model improvements.
A calm, vigilant approach helps protect your savings through uncertain times. Join Briefs Finance CEO Jaspreet Singh on September 29th for a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, where he shows how we're spotting investment opportunities as the dollar falls. Save your spot.
Skeptics of the slowdown movement are getting louder. David Sacks, the former White House crypto czar who now co-chairs the President's Council of Advisors on Science and Technology, told the tech giants to "stop pretending the motivation to slow down is purely altruistic." Emil Michael, serving in the role of the Defense Department's research and engineering undersecretary, warned of a "coordinated campaign" of fearmongering. Meanwhile, OpenAI said Monday it would hold back GPT-6.1 Astra after determining the model fell short of the company's safety standards. The following day, Anthropic reportedly cautioned potential IPO investors that the technology could pose "catastrophic or existential risks to humanity."
Mistral builds models too, but it favors bespoke deployments, working with individual companies to tailor tools and plug them into existing workflows. Unlike some competitors, it isn't planning a slowdown. Mensch said U.S. labs' lead is "not extremely large," adding that Mistral's next model should narrow the gap "very significantly." He also said the company is raising the capital to scale the compute needed for larger, more powerful models "in order for us to own our own destiny." Earlier this month, Mistral closed a 3 billion euros ($3.5 billion) round of new financing with Samsung, the memory-chip behemoth, taking the lead as it positions itself as a European alternative to OpenAI and Anthropic.
Keeping a steady plan lets your investments grow while risks are monitored. Our CEO Jaspreet Singh is hosting a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, on September 29th. Sign up free to join him live.
