The bosses of artificial intelligence firm Anthropic are set to meet senior White House officials on Monday after the US government forced the company to block public access to its most powerful AI models yet — just days after they were released.
The meeting in Washington DC between chief executive Dario Amodei and US commerce secretary Howard Lutnick comes after Anthropic suspended access to its latest AI tools, named Fable 5 and Mythos 5, on Friday. The company had previously described them as “too powerful” for the public.
“Anthropic blocked its powerful AI models after US government raised national security concerns, prompting a high-stakes White House meeting.”
The decision followed a directive from the US government prohibiting Anthropic from allowing any foreign national to use the technology. The government said it had “become aware” of a potential “jailbreak” — a vulnerability that could allow someone to make the AI do something it was not designed to do. Anthropic said it had received only “verbal evidence” of the alleged flaw.
Both Fable 5 and Mythos 5 are versions of Claude Mythos, an AI model that caused serious concern among technology, finance and government leaders when it was first previewed privately in April. At the time, Anthropic said 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 attention on Mythos was warranted, in part, because “it’s the unknown, unknown”.
When Anthropic released Fable 5 to the public last week, it acknowledged: “Releasing a model this capable comes with risks.” It added: “Fable’s capabilities exceed those of any model we’ve ever made generally available.” Within days, the government stepped in.
The showdown is the latest in a fractious relationship between Anthropic and the US government. The company is already suing the US Department of Defense over a dispute about how its models can be used. But tensions had appeared to ease a few weeks ago when Anthropic met senior White House officials in what was described as a “productive” meeting.
Now, Monday’s meeting with the commerce department is expected to include more documentation of the alleged jailbreak vulnerability, according to a person familiar with the matter. It remains unclear whether Anthropic will be allowed to make Fable 5 and Mythos 5 accessible again.
Meanwhile, dozens of tech leaders and cybersecurity executives — including staff from Nvidia, Zoom, Mercedes-Benz, and former US government and Google security staff — have signed an open letter urging the government to allow the models’ release.
Anthropic is also preparing to go public soon, with its private valuation nearing $1tn (£747bn). The company’s ability to demonstrate ever-broader AI capabilities underlines its value to investors.
On Tuesday, Anthropic said the roughly 150 groups that had previewed Mythos would now get access to Claude Mythos 5, which lacks certain limitations on cybersecurity and biology depending on the organisation’s uses. The company said the updated access was limited to a “small group of cyberdefenders and infrastructure providers” but that it intended to expand it through a “broader trusted access program”.
Anthropic co-founder Jack Clark told BBC Newsnight last week that AI capabilities were expanding so rapidly that 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.