EU seeks to ‘intensify’ talks with U.S. on AI models amid Mythos concerns
The European Union is considering “intensifying” talks with the US administration on the most advanced artificial intelligence 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, sparking a wave of concern about AI-powered cyberattacks. Same month, Wall Street Journal reported The White House has opposed Anthropic’s plan to expand access to its powerful Mythos model.
On Thursday, Anthropic said it expects to offer “Mythos-class” models to customers in the coming weeks.
The White House says it is working closely with AI labs to strike a balance between innovation and security as the United States seeks to maintain its lead over China in the global AI race.
Anthropic initially rolled out the model in preview to a select group of companies and organizations as part of an initiative called Project Glasswing. But the AI lab has yet to grant preview access for review to the EU, its AI office, or any non-US government body other than the AI Security Institute in the UK.
WSJ reported that Anthropic proposed allowing nearly 70 companies and organizations to use Mythos, a move that administration officials opposed due to security concerns.
“Cybersecurity is a common priority and we agreed to mutually recognize our relevant standards in this area,” Commission spokesman Thomas Regnier told CNBC, referring to talks between the Commission and the US administration. he said.
“In parallel, we are expanding technical discussions with Anthropic and other model developers who have reported their latest models to the AI office.”
During discussions about preview access to Mythos, Anthropic told the Commission that the bloc should first seek permission from the US administration, a person familiar with discussions between the EU and Anthropic told CNBC. The person requested to remain anonymous because they are not allowed to speak publicly about confidential discussions.
The source added that Anthropic told the Commission that the White House was not specifically opposed to sharing Mythos with the EU, but that the administration was more generally opposed to the company sharing it with non-US governments.
The White House press office directed CNBC to inform Treasury Secretary Scott Bessent of his comments when approached for comment on Thursday. Bessent said the US administration is working “very closely” with artificial intelligence laboratories.
“They were perfect partners, and we will find a solution that solves maximum computation,” Bessent said, adding that the United States is the world’s AI leader and China is second. “We want to make sure we maintain that lead. So we’re working on the exact calculation between innovation and safety, and we want to optimize for that.”
Anthropic said Thursday that models of Mythos’ capability require strong cybersecurity measures before they can be released publicly. “We are making rapid progress in developing these safety measures and hope to be able to offer 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 exposed by the company’s Mythos model until Chinese AI catches up.
Anthropic declined to comment.
