Supporters close to Anthropic and the US government have reportedly raised security concerns with Mythos and Fable.
Countries around the world are reeling from the consequences of the sudden shutdown of two of Anthropic’s most capable AI models following an order from the US government late Friday (June 12).
Now, the AI giant’s top technology staff are reportedly in Washington to meet with White House officials in hopes of resolving their second major dispute in months.
In a surprise statement on Friday, Anthropic said the US government issued an export control order, ordering the company to stop access to Claude Fable 5 and Claude Mythos 5 “by any foreign person, whether inside or outside the US, including foreign employees of Anthropic”.
The order effectively halted the use of the company’s leading cybersecurity model Mythos and its banned version, Fable, released last week.
The letter, sent to Anthropic at 5:21 pm EST (10:21 pm Dublin Time), cited national security concerns but did not provide specifics, the company said.
“Our understanding is that the government believes that it already knows how to get through, or ‘Nsumansumane 5’ to break into the prison.
“This vulnerability appears to be relatively simple, and we’ve found that some publicly available models are able to detect it without requiring a bypass.
“We have reviewed the report which we believe is the basis of the government directive and confirmed that the level of capability shown there is widely available in other models,” the statement read.
David Sacks, the former ‘czar’ of AI and chairman of the President’s Council of Advisors on Science and Technology said that a “very reliable partner” of Anthropic and the US government that was “testing Fable” told officials about a possible prison break by Fable’s guards.
“In their blog post, Anthropic defended its decision by saying that the jailbreak is not serious. That’s not what our trusted partners and the USG believe; and it’s not the kind of reductive language that fits Anthropic’s brand as an AI security company,” Sacks wrote on the X blog.
And Pentagon chief information officer Kirsten Davies said the move prioritizes national security. “Some things are more important than revenue cycles, clickbait and pre-IPO valuations,” Davies said.
Meanwhile, several news publications reported on Saturday (June 13) that Amazon CEO Andy Jassy was among the technology leaders who raised concerns with the US government about potential security concerns in Anthropic models, which led to a ban on export controls. Amazon is a major investor in Anthropic.
An Amazon company spokesperson told TechCrunch that “it is unusual for governments to seek our advice on potential security risks”.
The European Commission is among the external regions facing the consequences of the sudden suspension of the latest Anthropic models. Several countries in the bloc, as well as the UK, have been in discussions with the company about Mythos and its potential applications in securing their systems.
In a statement on Sunday 14 June, the Commission said it is analyzing the effects of the export control order.
“We are seeing a new generation of highly capable AI models coming to the market. These models provide important benefits, including cyber protection, but they also raise serious concerns about cyber security that need to be addressed,” said a European Commission spokesperson.
“We believe that the emergency measures taken in this way should not discriminate against our partners.
“This development is a further illustration of why Europe needs to strengthen its technological sovereignty…We are very much looking at the practical consequences of this for European users of these services.”
In the same statement, Canadian prime minister Mark Carney said the incident shows the dangers of over-reliance on a small number of powerful AI tools.
This is the second major dispute Anthropic has with the US government, which could have serious consequences for the AI giant as it prepares to go public in an estimated $1trn-plus listing.
The company is currently embroiled in a legal dispute with the Trump administration following a separate incident earlier this year after Anthropic refused to change its AI protections for use by the US government.
Anthropic said its designation in late February of ‘supply chain risk’ by the US government was not legal. Its latest statement last week made the same connection, saying the actions of the Trump administration are inconsistent with the principles of transparency, fairness and facts.
The tension between Anthropic and the US government comes as the dominance of eye technology in AI for their contemporaries.
Don’t miss out on the information you need to succeed. Sign up for Daily BriefSilicon Republic’s digest of must-know sci-tech news.
