AIToday

Anthropic's Mythos models still offline after 2 weeks of talks with Trump administration

The Verge AI23h ago5 min read
Anthropic's Mythos models still offline after 2 weeks of talks with Trump administration

Key takeaway

Anthropic's Mythos-class AI models have been offline for two weeks after a Trump administration export control order, with no resolution in sight despite intense negotiations. The shutdown threatens Anthropic's path to profitability and upcoming IPO, while signaling to other US AI companies and foreign governments that America is willing to restrict access to its most powerful AI systems on security grounds.

Summaries like this, in your inbox every morning.

Sign up free →

3 Key Points

  • What happened

    Anthropic took its Mythos-class and Fable 5 models offline following a June 12th Trump administration export control order banning access by any foreign national due to security concerns. Two weeks of high-intensity negotiations have produced no resolution, and the company says there is no news to share.

  • Why it matters

    Mythos models were supposed to boost Anthropic's revenue ahead of an IPO, with input tokens priced at double the cost of lower-powered alternatives. The prolonged shutdown also signals the US government is willing to lock down American AI systems it deems risky, creating uncertainty for other companies like OpenAI, Google, and Microsoft—and potentially opening space for non-American AI systems globally.

  • What to watch

    The impasse stems partly from a lack of clear framework for applying export controls to AI systems, and a reported vulnerability in Fable 5 (flagged by Amazon CEO Andy Jassy) that some cybersecurity experts say was overblown. Anthropic cofounder Tom Brown and public policy chief Sarah Heck are now leading talks, but progress remains unclear.

FAQ

Why did the Trump administration order the models offline?
The June 12th export control order cited security concerns and banned access by any foreign national to Mythos 5 and Fable 5. The order was reportedly triggered after Amazon CEO Andy Jassy flagged a method for seemingly breaking Fable 5's guardrails that prevent it from finding exploitable security holes.
Why hasn't this been resolved in two weeks?
There is no clear framework for applying export controls to AI systems like there is for traditional dual-use manufacturing products. The process normally unfolds over months or years, but was compressed into a few days in this case, leaving both sides trying to figure out how to apply the rules from first principles.
What does this mean for other US AI companies?
OpenAI, Google, and Microsoft have models that may pose similar risks to Mythos, and the Trump administration reportedly asked OpenAI to delay the release of GPT-5.6 over security concerns. The situation has created a power vacuum in the global AI market and prompted countries to call for non-American AI alternatives.

Discussion

No discussion yet for this article

Stay ahead with AI news

Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.

Get Started Free

Free · takes 30 seconds · unsubscribe anytime

1 minute a day. The AI essentials.

200+ sources · Email / LINE / Slack

Get it free →