White House gives Anthropic 90-minute deadline to withdraw AI models

Anthropic, the AI safety company behind the Claude chatbot, faced a stark choice on June 13, 2026: restrict access to its most powerful AI models within 90 minutes or face the consequences. The company chose a third option nobody expected. It pulled the models entirely.

The White House ultimatum targeted Anthropic’s Fable 5 and Mythos 5 models, demanding the company limit access exclusively to US citizens. Rather than scramble to implement nationality-based restrictions in the time it takes to watch a movie, Anthropic withdrew both models from availability altogether.

A confrontation months in the making

This wasn’t a bolt from the blue. The Trump administration and Anthropic had been on a collision course for months, with tensions escalating throughout early 2026 over questions of model oversight and who gets to decide how frontier AI is deployed.

The friction intensified to the point where the Pentagon placed Anthropic on a blacklist, a move that would have been unthinkable even a year earlier for a company that had positioned itself as the “responsible AI” player in Silicon Valley.

On June 2, 2026, the administration issued an executive order establishing frameworks for evaluating what it termed “covered frontier models” based on their cyber capabilities.

Eleven days later, the export controls dropped on Anthropic’s doorstep. The core concern, according to the administration, was potential foreign access to these advanced systems, with China being the specific worry.

From voluntary guidelines to forced compliance

This represents a fundamental shift in how Washington approaches AI regulation. The previous posture, even under this same administration, leaned heavily on voluntary commitments from AI companies. That era appears to be over.

Giving a company 90 minutes to overhaul its entire access infrastructure isn’t a negotiation. It’s a demand. And the fact that Anthropic chose total withdrawal over rushed compliance tells you something about how technically complex, or politically fraught, the alternative was.

The executive order from June 2 laid the groundwork by creating evaluation criteria specifically focused on cyber capabilities of frontier models. This suggests the administration views advanced AI not primarily as a commercial product or research tool, but as a potential weapon requiring the same kind of export control regime applied to advanced semiconductors and military technology.

What this means for investors and the AI industry

For crypto and blockchain-based AI projects, decentralized AI platforms have marketed themselves partly on the idea that no single government can control access to their models. This incident will likely fuel that narrative, potentially driving interest and capital toward decentralized alternatives that are architecturally resistant to this kind of top-down intervention.

The broader concern for tech investors is that regulatory unpredictability is now a first-order risk factor for AI companies. Building a business on frontier AI means accepting that your core product could become subject to export controls overnight, with compliance timelines measured in minutes rather than months.

The Pentagon blacklist against Anthropic adds another dimension. Defense contracts represent a significant revenue stream for AI companies, and being locked out of that market while simultaneously facing export restrictions is a double blow that could reshape how investors value companies perceived as being on the wrong side of Washington’s priorities.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.