Mistral CEO Arthur Mensch has accused the world’s leading artificial intelligence companies of using the debate over the safety of their technology to cover up their own “negligence.”
“The discussions we’ve seen in the U.S. were a cover for the inaction of some of our competitors,” Mensch told CNBC’s Annette Weisbach, adding that systems need to be created to contain AI agents.
Mensch is not alone in criticizing the debate over the safety of AI in the United States. Former White House cryptocurrency czar David Sachs, who currently co-chairs the President’s Council of Advisors on Science and Technology, urged tech giants to “stop acting as if their motivation to slow down is purely altruistic.”
Emile Michael, the undersecretary of defense for research and engineering, also warned of a “coordinated campaign” of fear-mongering.
CNBC has reached out to OpenAI and Anthropic for comment.
Incidents involving rogue AI agents are on the rise globally, with the latest similar report coming from Australia, where an agent developed by OpenAI hacked a government website.
In July, Anthropic described three instances in which its Claude models accessed the Internet during evaluations and compromised live systems at three different organizations.
“If you give them a lot of tools, those systems (AI agents) become very dynamic, so they can do things you don’t expect,” Mensch told CNBC.
“We need to have proper monitoring in place and we provide such monitoring systems to the companies we work with,” he said.
AI slowdown
Debate over the safety of AI intensified earlier this month after an AI researcher left Anthropic and said the company and rival OpenAI were “risking their lives” and were building technology that had a more than 10% chance of killing all humans.
Anthropic co-founder and CEO Dario Amodei later published an essay urging companies to slow down the pace of improving their AI models.
OpenAI announced on Monday that it has decided not to release its next AI model, GPT-6.1 Astra, because it has determined that it does not adequately meet the company’s safety standards. Meanwhile, the following day, it was reported that Anthropic had warned potential investors in its IPO that the technology could pose “catastrophic or existential risks to humanity.”
Mistral also develops AI models, but takes a different approach than OpenAI and Anthropic, working with individual companies to create custom AI tools tailored to their needs and integrating them into their businesses.
But unlike its competitors, Mistral has no plans to delay development of advanced models.
Mensch said the gap with U.S. labs is “not extremely large,” adding that the company is confident its next-generation model will close the gap “very significantly.”
“We have raised the funding needed to scale the computing needed to train bigger and more powerful models to decide our own destiny,” he said.
Earlier this month, Mistral raised 3 billion euros ($3.5 billion) in new funding led by memory chip giant Samsung, as it seeks to position itself as a European alternative to rivals such as OpenAI and Anthropic.
