Anthropic Withdraws Newest AI Model Following US Government Order

AI company Anthropic has disabled its most powerful AI models, Claude Fable 5 and Mythos 5, following orders from U.S. authorities. According to reports from the BBC, CNN, and the New York Times, the shutdown follows a government mandate requiring the company to block foreign citizens from accessing the Fable 5 model due to security concerns.

Anthropic stated on its website that the company had to deactivate both Fable 5 and Mythos 5 for all customers to ensure compliance with the order. The company previously described the Fable 5 tool as “too powerful,” noting its ability to exploit or hack data systems.

Why did the U.S. government order the shutdown?

U.S. security authorities identified a “jailbreaking” method that allows users to bypass the model’s security mechanisms. While Anthropic claims the authorities did not point to specific concerns, the company acknowledged a demonstration of the technique used to find a small number of known vulnerabilities.

Anthropic argued in a press release that these vulnerabilities are simple and present in other publicly available AI models. The company maintains that other tools could find the same weaknesses without needing to bypass security measures.

Did You Know? The British AI Security Institute found in its testing that the Claude Fable 5 model successfully broke through defenses and systems 73 percent of the time.

What are the security risks associated with Claude Fable 5?

The model’s capabilities have drawn scrutiny from leaders in technology, finance, and government. Gina Neff, a professor of responsible AI at Queen Mary University of London, told the BBC that this represents a “shift in pace” for cybersecurity capabilities.

Neff warned that restricting access could hinder the safe testing and further development of such systems. Anthropic had previously granted early access to a few organizations in April because the tool was advanced enough to be potentially dangerous.

Expert Insight: Samantha Carter suggests that the clash between Anthropic’s “too powerful” branding and government security mandates highlights a growing tension. The stakes involve a trade-off where the pursuit of cutting-edge AI capabilities may directly conflict with established national security protocols regarding software vulnerabilities.

How does the legal dispute with the Trump administration affect Anthropic?

Anthropic is currently in a legal battle with the Trump administration over a directive intended to stop government agencies from using the company’s AI tools. Former Defense Secretary Pete Hegseth labeled the company a “supply chain risk,” a term typically reserved for companies in hostile nations.

Anthropic disables top AI models after US foreign access order

The company has sued the Pentagon in response to these actions. A U.S. judge has ruled that the directive cannot be enforced while the lawsuit is ongoing, which allows defense partners and government agencies to continue using Anthropic’s tools.

What may happen next?

The future of the Fable 5 model may depend on whether Anthropic can satisfy government security requirements regarding foreign access. Further legal rulings in the case against the Pentagon could determine if the “supply chain risk” label remains, which may influence how the U.S. government integrates AI into public sector operations.

If security vulnerabilities are not resolved to the satisfaction of U.S. authorities, the company could face continued restrictions on its most advanced releases.

Frequently Asked Questions

Which AI models did Anthropic deactivate?
The company deactivated Claude Fable 5 and Mythos 5.

What is “jailbreaking” in the context of this story?
It is a method used to bypass security mechanisms in software or devices to gain access to more functions.

Why did some critics dismiss Anthropic’s claims about the model’s power?
Some critics, according to the BBC, viewed the claim that the model was “too powerful to be released” as exaggerated marketing.

How should governments balance the drive for AI innovation with the need to prevent national security risks?