Two weeks after the ban forced Anthropic to pull its powerful cybersecurity models Mythos 5 and Fable 5 from the market, the Trump administration has softened its stance. Commerce Secretary Howard Lutnick notified Anthropic that Mythos 5 can be redeployed to more than 100 specific US organizations, including government agencies and private companies. The most notable change is that non-American employees of these entities are now allowed access, reversing the original prohibition that barred non-US personnel from using the models.
Details of Lutnick's directive
According to a letter seen by Semafor, Lutnick wrote to Anthropic's chief compute officer Tom Brown on Friday: "I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model." The directive does not address Fable 5, a version of Mythos 5 released a few days before the ban which included additional protections that were reportedly bypassed easily by security researchers. Anthropic did not immediately comment but acknowledged progress on X: "Since June 12, we have been working closely with the US government to restore access to Claude Mythos 5 and Fable 5. Today, the government notified us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical infrastructure."
Sponsored Protocol
Implications for national security and the AI industry
This decision represents a compromise between national security concerns and technological innovation. Mythos 5 was blocked due to fears that it could be exploited by adversaries if accessed by non-US personnel. However, pressure from companies relying on these models to protect their networks prompted the administration to reconsider. The limited access to over 100 entities serves as a test to evaluate the effectiveness of implemented safeguards. Whether Fable 5 will receive similar treatment remains uncertain. This case raises questions about the delicate balance between AI export controls and the competitiveness of American businesses, a topic also explored in articles on US government restrictions on OpenAI models. Furthermore, the European regulatory framework with the EU AI Act demonstrates different approaches to AI regulation.
Sponsored Protocol
The future of cybersecurity with models like Mythos 5
With this controlled reintroduction, Anthropic can demonstrate that AI models can be used safely to defend critical infrastructure. Continued collaboration with the US government may lead to a more stable regulatory environment, enabling advanced tools like Mythos 5 to be deployed at scale without compromising national security. For more on AI's role in cybersecurity, see the Wikipedia page on AI for cybersecurity.
Sponsored Protocol
