The order prohibits all foreign nationals - including Anthropic’s own staff - from using the models, forcing the company to shut down access for all customers.
The company says the authorities have not specifically stated what their concerns are. However, Anthropic says authorities believe they have become aware of a method to bypass the security barriers of the Fable 5 model that could facilitate hacking.
Fable 5 is a locked version of Mythos 5, an advanced model withheld from the public because it is highly effective at finding vulnerabilities in program code.
Anthropic disagrees with the authorities' risk analysis, stressing that Fable 5 does not provide hackers with unique tools. The company warns such measures could halt development across the industry.
This comes amid a legal battle in which Anthropic has refused the Trump administration's requests to use its technology for mass surveillance and autonomous weapons, leading the Pentagon to break its contract with the company.
Claude Mythos is an AI model developed by American company Anthropic. The model can write and understand programming code, and according to the company, is better than almost all humans at finding and exploiting vulnerabilities.
How to use it:
1. The model is allowed to read the program's source code on a computer isolated from the internet.
2. Mythos develops a hypothesis about possible vulnerabilities.
3. The hypothesis is tested to confirm or reject the suspicions; if needed, it is adjusted and tested again.
4. Once Mythos has found several smaller vulnerabilities, it tries to chain them together to create a more powerful attack.
5. The model writes a bug report.
Source: Anthropic
