Trump administration bars foreign access to Anthropic's Mythos and Fable models
Cover image from nypost.com, which was analyzed for this article
Anthropic disabled access to top models like Mythos for foreign nationals after a Trump administration national security directive. The move follows broader US efforts to control advanced AI exports. Tech reporting covers company responses and White House reactions.
PoliticalOS
Monday, June 15, 2026 — Tech
The U.S. government has asserted direct control over who may use frontier AI models developed by American companies, citing national-security risks that include a reported jailbreak and potential foreign access. This action immediately affects Anthropic’s operations and creates openings for Chinese developers while leaving the long-term balance between security restrictions and commercial innovation unresolved.
What outlets missed
The formal Commerce Department export control directive issued June 12 that directly compelled the takedown received little technical detail across coverage. No outlet provided independent confirmation of the jailbreak technique or the identity of the additional companies that reportedly alerted officials. The impact on Anthropic’s own foreign-national employees, who can no longer work on the restricted models, and the precise scope of the China-linked access allegation were mentioned only in passing or not at all.
The Trump administration imposed export controls on Anthropic’s most advanced AI models, Mythos and Fable, after receiving reports that researchers at Amazon had bypassed Fable’s safety guardrails. The restrictions, issued Friday, prohibit use by foreign nationals inside or outside the United States and prompted Anthropic to take both models offline entirely.
The controls followed a reported jailbreak identified by Amazon CEO Andy Jassy and conveyed to Treasury Secretary Scott Bessent. Administration officials stated that nearly half a dozen companies had raised similar concerns. A separate report indicated that a China-linked entity had accessed Mythos despite existing company prohibitions on Chinese use. Anthropic described the Fable flaw as narrow and not unique to its systems, and said it had received prior government approval to release the model.
Anthropic filed confidentially for an IPO earlier this month and has publicly warned that its frontier models carry significant risks, including potential for devastating cyberattacks and a 25 percent chance that advanced AI could destroy humanity. The company stated it is working to address the administration’s concerns so the models can return to wider availability. White House AI adviser David Sacks said the goal remains remediation followed by lifting of the controls.
Chinese AI developer Zhipu announced the open-source release of its GLM-5.2 model the same day, framing the move as a contrast to U.S. restrictions. Its Hong Kong-listed shares rose 33 percent. Analysts at JPMorgan and Bank of America raised price targets on Zhipu while downgrading rival MiniMax.
Meetings between Anthropic technical staff and the Commerce Department, CIA, and White House science adviser are scheduled for Monday. The episode leaves unresolved how the United States will balance national-security controls on frontier models against the operational needs of companies whose workforces include foreign nationals and whose growth plans depend on broad model access.
More in Technology
SpaceX Moves to Acquire Cursor Parent for $60 Billion After IPO
SpaceX agreed to buy AI coding startup Cursor's parent Anysphere for $60 billion in stock, marking a major enterprise push shortly after its IPO and drawing tech market attention.
AI agents, export curbs and deepfake ads test tech's next phase
Reports covered expanding AI agent capabilities, regulatory scrutiny, and workforce impacts, including Qualcomm's comments on AI replacing apps and broader Big Tech moves.
Meta Says Threads Reached 500 Million Monthly Users
Meta's Threads platform hit half a billion monthly active users with new personalization features, highlighting social media competition and growth trends.
UK to Bar Under-16s From Social Media Platforms in 2027
The UK government announced a sweeping ban on social media apps including TikTok and YouTube for children under 16 to protect childhood development. The policy drew international attention and comparisons to other nations. Coverage includes reactions from tech firms and parents.
The Compass
You just read five takes on one story.
What's your take? Find your political shape in a few minutes.
Take the test