Washington’s AI Blackout: Anthropic Forced to Pull Plug on Fable 5 and Mythos 5

A surprise government directive has triggered a sudden global freeze on Anthropic’s most advanced models, raising urgent questions about the future of AI export controls.

For the developers at Anthropic, the celebration of their latest release was short-lived. Just three days after the public launch of Fable 5, the firm was forced into a sudden, total blackout of its most powerful software. At 5:21 p.m. on Friday, a directive from U.S. Commerce Secretary Howard Lutnick arrived at their headquarters, mandating an immediate suspension of access to both Fable 5 and the ultra-sophisticated Mythos 5. The order is sweeping; it prohibits all foreign nationals—including those employed by the company itself—from accessing the technology, effectively forcing the firm to pull the plug on its customers worldwide to remain in compliance.

The Security Standoff

The government has remained tight-lipped regarding the specific intelligence behind this move, but the industry consensus points to a single, contentious issue: "jailbreaking." While Fable 5 was designed as a secure, locked-down iteration of the more potent Mythos 5, federal officials reportedly believe that a vulnerability exists that could allow the model to assist in malicious hacking. Mythos 5, the unrestricted engine behind the platform, had already been kept on a tight leash, released only to a select group of corporate partners due to its unsettling ability to identify deep-seated software vulnerabilities.

Anthropic is pushing back hard against the narrative. The firm says it has thoroughly investigated the alleged bypass methods and insists that Fable 5 offers no capabilities to bad actors that aren't already present in other publicly available models. They argue that if regulators applied this standard of "narrow potential risk" to the rest of the industry, the development of frontier models would grind to a complete halt.

The Bigger Picture

This clash is not an isolated incident; it marks a significant escalation in the ongoing friction between the Trump administration and the tech sector. Anthropic has spent months in a tense legal standoff with the government, largely over the company’s refusal to allow its infrastructure to be utilized for mass surveillance or the development of autonomous weaponry. The fallout has been tangible, with the Pentagon already moving to sever existing contracts with the firm.

For the broader market, this episode highlights a jarring new reality. As frontier models become more powerful, they are being treated less like software and more like munitions, subject to the heavy hand of national security export controls. The precedent set here is chilling: a single government letter can now effectively vaporize a product that has already been deployed to millions of users. As the line between innovation and national security continues to blur, companies are learning that even the most robust security safeguards might not be enough to satisfy Washington’s growing demand for total control.