Anthropic remains locked in a two-week standoff with the White House after taking Mythos-class AI models offline under a federal export order. Key Points: Anthropic has kept Mythos 5 and Fabl
Anthropic remains locked in a two-week standoff with the White House after taking Mythos-class AI models offline under a federal export order.
- Anthropic has kept Mythos 5 and Fable 5 offline since the Jun. 12 order.
- The order blocks access by any foreign national, including non-U.S. employees.
- The dispute could reshape how Washington treats frontier AI releases.
The standoff began after the Trump administration ordered Anthropic to suspend access to Mythos 5 and Fable 5 for any foreign national, citing security concerns tied to the systems’ advanced capabilities.
The order covered non-U.S. citizens inside and outside the country, including Anthropic employees, leaving the company with few practical options beyond keeping the models offline.
The Verge reported that Anthropic sent executives to Washington after the order, but the company has declined repeated comment on the talks, saying there was no update to share.
The impasse appears tied to the lack of a settled export-control process for AI systems, since dual-use products are usually reviewed before market release under clearer manufacturing rules.
A person familiar with the talks told The Verge that the Commerce Department tested Fable 5 before release and raised no objections, before a later report about a guardrail bypass changed the timeline.
Also Read:USDT Briefly Dethrones Ethereum As Crypto’s No. 2 Asset
Katie Moussouris, founder and CEO of Luta Security, reviewed the reported Fable 5 issue at Anthropic’s request and argued that the vulnerability had been overstated.
She wrote that asking an AI system to fix bugs, explain the fix and test the patch is normal defensive security work, not a reason for a sweeping shutdown.
The delay has raised the stakes for Anthropic’s business, because Mythos-class models were expected to support revenue before a possible IPO and help pay for compute commitments, including a reported SpaceX data center deal.
The broader concern is that other U.S. AI companies could face similar controls as their models approach Mythos-level capabilities, including OpenAI, Google and Microsoft. The dispute follows months in which Washington sought to reduce AI regulation, making the Mythos order a sharp reversal and a potential template for future frontier-model restrictions.
Read Next:Techdollar Raises $3M To Let Startup Workers Cash In Without Selling
