US Ask Anthropic To Suspend Access To Fable 5, Mythos 5: What Does This Mean
The United States government has taken the unprecedented step of ordering artificial intelligence firm Anthropic to immediately freeze access to its two flagship AI systems, Fable 5 and Mythos 5. Citing national security powers, Washington has issued a sweeping export control directive that bans any foreign citizen, whether located globally or inside America, from using the software. Crucially, the restriction is so strict that it even prohibits Anthropic’s own foreign national employees from accessing their company’s technology. Because the tech firm cannot easily filter users by nationality in real time, it has been forced to completely switch off both systems for all customers worldwide.
The abrupt shutdown has sent shockwaves through the technology sector, leaving hundreds of millions of users without access. Anthropic revealed it received the legal directive late on Friday afternoon and had no choice but to comply. While other older Anthropic models remain online, the sudden recall of its most advanced tools marks a dramatic escalation in how governments regulate commercial artificial intelligence.
At the heart of the dispute is a security concept known as “jailbreaking.” This happens when a user clever phrases a prompt to trick an AI system into bypassing its built-in safety rules, potentially forcing it to generate harmful content or assist in cyberattacks. The US government fears that a newly discovered loophole allows users to break through the digital guardrails of Fable 5, creating a pressing national security threat.
However, Anthropic has hit back at the decision, arguing that the government has overreacted to a minor issue. According to the company, the state’s anxieties stem from a verbal report about a “narrow” loophole. In this specific scenario, the AI was successfully tricked into reading a piece of software code and identifying its flaws. Anthropic insists that this level of capability is not unique or inherently dangerous. They argue that the trick merely uncovered minor, previously known vulnerabilities that could easily be found using rival, publicly available systems, such as OpenAI’s GPT-5.5. Furthermore, the company points out that IT experts and cybersecurity defenders use AI in this exact way every single day to patch security flaws and keep the internet safe.
Before launching Fable 5, Anthropic spent thousands of hours stress-testing the model alongside government agencies, including the UK Artificial Intelligence Safety Institute. The company maintains that while its safeguards are among the toughest in the industry, “perfect resistance” to trickery is technologically impossible. Instead, they rely on a layered defense strategy, closely monitoring user data to spot and shut down malicious activity as it happens.
Anthropic has expressed deep frustration over the lack of transparency, noting that the government’s official letter failed to provide specific technical details of the alleged threat. The tech firm warned that if Washington continues to recall major commercial products over minor, localized safety bypasses, it will effectively paralyse the entire AI industry and halt the rollout of new technology.
While Anthropic supports the idea that governments should have the power to block truly dangerous technology, it argues that such decisions must be fair, transparent, and rooted in clear scientific facts rather than sudden panic. The company has apologised to its global customer base for the massive disruption, branding the incident a profound misunderstanding. Executives are currently locked in urgent talks with US officials, working to clarify the situation and restore access to Fable 5 and Mythos 5 as quickly as possible.
