How the world's top AI models were revived

After a twenty-day standoff with federal regulators, Anthropic's latest AI models—Mythos and Fable—were restored to service this week. According to Axios, the shutdown began when Amazon, Anthropic's partner and investor, flagged what it called a "jailbreaking" issue—essentially a technical vulnerability that could bypass the models' safety safeguards. Amazon escalated the concern to the Trump administration in mid-June, prompting Commerce Secretary Howard Lutnick to call Anthropic CEO Dario Amodei with an ultimatum: fix the problem, or face sweeping export controls. The models went dark immediately. Over the next three weeks, Anthropic deployed engineers to Washington D.C. to work with federal agencies including the NSA and the White House Office of Science and Technology Policy. Technical teams scrutinized the fixes line-by-line. By July first, multiple agency heads signed off and the models returned to service. Cybersecurity experts later disputed Amazon's framing, noting that other leading AI labs face similar vulnerabilities. The broader question remains: how will the government approve future models from OpenAI and Google, and when will allied nations gain access to these systems?

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton