US government forces Anthropic to disable its most advanced AI models worldwide over minor jailbreak fears

The US government just launched an unprecedented regulatory strike on the commercial AI market. Following an abrupt export control directive received at 5:21pm ET, Anthropic has been forced to completely disable its newly launched Fable 5 and Mythos 5 models for all users globally. The sudden shutdown impacts hundreds of millions of people and demonstrates that federal regulators are now willing to weaponize national security authorities to recall major tech products overnight. The immediate consequence is a total commercial freeze on Anthropic’s flagship offerings, causing massive disruptions for enterprise customers and casting a dark cloud over the financial viability of future frontier model deployments.

The devastating suspension stems from an export control order aimed at preventing any foreign national including Anthropic's own non-US employees from accessing the systems. Because it is functionally impossible to instantaneously verify the citizenship of every single user, the company was forced to pull the plug entirely to ensure compliance. Anthropic announced the shutdown in a stark public statement and an accompanying X post, apologizing for the disruption while fiercely defending its safety testing protocols.

According to the company, the government’s directive is based on a fundamental misunderstanding of artificial intelligence security. Officials cited a method of bypassing or "jailbreaking" the system to identify minor software flaws. However, Anthropic insists that they spent thousands of hours red-teaming the models with the US government and the UK’s AI Safety Institute prior to launch. The supposed vulnerabilities the government flagged are not a "universal jailbreak," but rather simple coding evaluations that rival systems, including OpenAI's GPT-5.5, can already perform.

The financial and operational implications of this federal intervention are staggering. Anthropic had already absorbed the real business costs of implementing a strict 30-day retention policy for customer data specifically to monitor and mitigate these types of risks. Now, despite acting proactively and operating under a defense-in-depth strategy, their primary revenue-driving models are frozen indefinitely.

By utilizing national security export controls on a commercial software platform, Washington has introduced a catastrophic new risk premium for tech investors and enterprise clients. If a minor, non-universal vulnerability is sufficient to instantly mandate a global recall, the entire artificial intelligence industry faces an existential operational threat. As Anthropic warned in its corporate response, "If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers." The company is scrambling to resolve the crisis, but the chilling effect on the AI business landscape is already locked in.