The US Commerce Department has lifted export controls on Anthropic’s advanced AI models following the company’s adoption of new security measures designed to prevent misuse, marking a crucial step toward resolving regulatory hurdles imposed during the Trump administration.
- Commerce Department ends export controls on Anthropic’s Fable 5 and Mythos 5.
- New safeguard routes sensitive queries to a less advanced AI model.
- Defense department still flags Anthropic as a supply chain risk.
What happened
Anthropic's highly advanced AI models, Fable 5 and Mythos 5, had been restricted under US export controls due to concerns raised by the Trump administration about potential security risks. These restrictions effectively took the models offline for broader user access, especially regarding certain sensitive capabilities related to cybersecurity and biology.
In response, Anthropic introduced a new security feature that detects and blocks risky user queries—those which might exploit vulnerabilities in the model’s behavior or circumvent existing controls. When blocked, the system automatically switches the request to a previous, less advanced model known as Opus 4.8. This targeted approach addressed the administration’s concerns, prompting the Commerce Department to remove restrictions and allow wider use of Anthropic’s latest AI models.
Why it matters
The removal of restrictions on Anthropic’s AI models signals a notable shift in US technology policy towards balancing innovation with national security. By requiring companies to implement proactive safeguards, the government aims to mitigate risks without stifling access to advanced AI tools critical for development.
However, this resolution applies primarily to Commerce Department oversight. Other government bodies, such as the Department of Defense, continue to regard Anthropic as a supply chain risk, which means the company’s regulatory challenges are not fully resolved. This ongoing scrutiny reflects the broader geopolitical and security debates surrounding AI technologies and their deployment.
What to watch next
Stakeholders should monitor whether Anthropic’s new security measures effectively prevent misuse and if any incidents arise that could prompt renewed government intervention. Continued evaluation by the Commerce Department’s Center for AI Standards and Innovation will be crucial to ensure compliance and safety.
Moreover, the Defense Department's designation of Anthropic as a supply chain risk remains a significant obstacle. Watch for potential policy shifts or negotiations that might influence this status, as well as industry responses that could impact AI model deployment and trust in the US and global markets.