The bosses of Anthropic are set to meet senior White House officials on Monday in Washington DC after the US government forced the artificial intelligence firm to block public access to its latest AI models, which the company itself had described as “too powerful”.
The meeting, between Anthropic chief executive Dario Amodei and commerce secretary Howard Lutnick, comes days after the company pulled its newly released tools – named Fable 5 and Mythos 5 – from public view. The government prohibited Anthropic from allowing any foreign national access to the technology, citing a potential “jailbreak” vulnerability that could allow the tool to be used for unintended purposes.
“Anthropic executives to meet White House after US blocks its 'too powerful' AI tool over national security concerns.”
Anthropic said on Friday it had only received “verbal evidence” of the purported vulnerability. The US government said it had “become aware” of the opening, though Anthropic has questioned the lack of documentation. The Monday meeting is expected to include more documentation of the alleged issue, according to a person familiar with the matter. It remains unclear whether Anthropic will be able to make Fable 5 and Mythos 5 accessible again after the talks.
The two models represent a version of Claude Mythos, an AI program that caused serious concerns among technology, finance, and government leaders when it was released privately in April for preview and testing. “Fable’s capabilities exceed those of any model we’ve ever made generally available,” Anthropic said last week, adding that “releasing a model this capable comes with risks”.
When Mythos was first shared with a small group of organisations, Anthropic warned the tool was so intelligent it could be dangerous because of its ability to exploit or hack computer systems. Canadian finance minister François-Philippe Champagne told the BBC in April that the attention on Mythos was warranted, in part, because “it’s the unknown, unknown”.
The clash is the latest between Anthropic and the US government. The company is suing the US Department of Defense over a dispute about how its models can be used. However, tensions appeared to ease a few weeks ago when Anthropic met senior White House officials in what was described as a “productive” meeting.
Despite the standoff, dozens of tech leaders and executives in the cyber security space have called on the US government to allow Anthropic to release the models to the public. In an open letter, security staff from Nvidia, Zoom, Mercedes-Benz, and former security staff for the US government and Google urged approval.
Anthropic is expected to become a public company soon, with its private valuation nearing $1tn (£747bn). The company also said the roughly 150 groups that had been given access to preview Mythos will now have access to Claude Mythos 5, which does not have limitations on cybersecurity or biology, depending on an organisation’s specific uses. To-date, those groups have reported finding more than 10,000 critical security flaws in their systems.
Anthropic co-founder Jack Clark last week told BBC Newsnight that the ability of AI tools was expanding so rapidly that the company thought there should be a way for the public to slow the technology’s advancement. “You want the option to be able to take your foot off the gas and put your foot on the brake,” he said.