The European Union is looking to “intensify” discussions with the U.S. administration on the most advanced AI models, including those with “cyber capabilities,” a Commission official told CNBC.
Anthropic’s powerful Mythos model, announced in April, sent governments and businesses into a frenzy, prompting a wave of concerns about AI-powered cyberattacks. That same month, the Wall Street Journal reported that the White House opposed Anthropic’s plan to expand access to its powerful Mythos model.
Anthropic on Thursday said it expects to bring “Mythos-class” models to customers in the coming weeks.
The White House says it’s working closely with AI labs to strike a balance between innovation and safety, as the U.S. seeks to preserve its lead over China in the global AI race.
Anthropic initially rolled out the model to a select group of companies and organisations in preview as part of an initiative called Project Glasswing. But the AI lab has yet to grant the EU, its AI office or any government organizations outside of the U.S., aside from the U.K.’s AI Security Institute, preview access to review it.
The WSJ reported that Anthropic had proposed letting roughly 70 additional companies and organizations use Mythos, with administration officials opposing the move because of security concerns.
“Cybersecurity is a shared priority and we have agreed to mutually recognise our respective standards in this area,” Thomas Regnier, Commission spokesperson, told CNBC, referring to talks between the Commission and the U.S. administration.
“In parallel we are now expanding technical discussions with Anthropic and other model developers which have notified their latest models to the AI office.”
During discussions about preview access to Mythos, Anthropic told the Commission that the bloc first has to ask the U.S. administration for permission, a person familiar with the discussions between the EU and Anthropic told CNBC. The person requested anonymity because they’re not permitted to speak publicly about the confidential discussions.
The source added that Anthropic had told the Commission that the White House was not opposed to sharing Mythos with the EU specifically, but the administration is more generally opposed to the company sharing it with non-U.S. governments.
The White House press office referred CNBC to briefing comments from Treasury Secretary Scott Bessent on Thursday when approached for comment. Bessent had said the U.S. administration is working “very closely” with AI labs.
“They’ve been excellent partners, and we are going to get a solution that solves for maximum calculus,” Bessent said, adding that the U.S. was the world’s AI leader with China second. “We want to make sure that we keep that lead. So, we are working on the exact calculus between innovation and safety, and we want to optimize for that.”
Models of Mythos’ capability require strong cyber safeguards before they can be generally released, Anthropic said on Thursday. “We’re making swift progress on developing these safeguards and expect to be able to bring Mythos-class models to all our customers in the coming weeks.”
CEO Dario Amodei warned earlier this month that there was a six- to 12-month window to patch tens of thousands of software vulnerabilities uncovered by the company’s Mythos model before Chinese AI catches up.
Anthropic declined to comment.
