AI company Anthropic announced that, at the request of the US government, it has stopped certain groups, including foreign nationals and some company employees, from using its most advanced products.
To comply with this request, the company has closed access to Claude’s latest models, Fable 5 and Mythos 5, for all customers. Anthropic stated that the decision is based on the government’s national security concerns but did not provide further details.
Access to the company’s newest models has been blocked
According to Anthropic, Fable 5 and Mythos 5 are no longer available to any customers. The company noted that these models are more powerful than previous versions in software engineering and visual understanding.
Anthropic had also acknowledged that its new technology could carry some risks of misuse. In April, it released a preview version to a few industry partners to assess the potential for the model to be used for purposes such as creating cyberattack tools.
Tensions with Washington are not new
In February, the US administration banned federal agencies from using Anthropic’s products. On the same day, Donald Trump described the company as a "radical leftist, woke" firm for not allowing military use.
At that time, Anthropic’s CEO Dario Amodei said the US government’s demands for mass surveillance and fully autonomous weapons conflicted with the company’s security policies.
Another aspect of this debate involved reports that the US military had previously used Anthropic’s Claude system for intelligence assessments, target identification, and war scenario simulations.
Security debate intensifies in the AI race
Anthropic positions itself as an ethical AI company. Its recent filing for an initial public offering also indicates that the technology race is continuing at full speed.
Meanwhile, countries like China and the United Arab Emirates are working to build their own AI infrastructures to reduce external dependence. The latest move against Anthropic has further highlighted the debates over security, access, and sovereignty in the AI competition.
