Anthropic announced it will abruptly disable its most advanced AI models, Fable 5 and Mythos 5, after receiving a directive from the U.S. government to restrict access for foreign nationals due to national security concerns. The decision comes despite the company stating it was only given limited and unclear evidence about the risks involved. This marks a significant escalation in how governments are beginning to regulate access to powerful AI systems, moving beyond hardware restrictions into direct control of software usage. The move affects users globally and highlights how quickly policy decisions can impact cutting-edge technologies.
The core concern raised by regulators involves a potential “jailbreak” method that could bypass safety guardrails built into the models. According to Anthropic, this issue is described as narrow and not universally applicable, meaning it may not represent a widespread or immediate threat. However, authorities appear to be acting cautiously, prioritizing worst-case scenarios over limited evidence. This has sparked disagreement from the company, which argues that such a response is disproportionate. The situation reflects the growing tension between innovation speed and safety oversight in the AI industry.
The conflict is also rooted in a broader breakdown in relations between Anthropic and the U.S. government. Earlier disputes emerged when the company refused to allow its AI systems to be used for domestic surveillance and autonomous weapons by the military. In response, the government placed Anthropic on a Pentagon contractor blacklist, further straining cooperation. This background adds a political dimension to the current decision, suggesting that regulatory actions may not be purely technical. It also raises questions about how much influence governments should have over private AI development.
At a deeper level, the case highlights increasing fears around advanced AI enabling sophisticated cyberattacks.