The Trump administration has ordered Anthropic to block foreign access to its advanced AI models, Fable 5 and Mythos 5, citing national security concerns. In response, Anthropic has disabled access to these models globally for all users.
Immediate Action & Core Facts
The U.S. Commerce Department directed Anthropic to restrict access to Fable 5 and Mythos 5 by foreign nationals, whether inside or outside the U.S. The move follows reports that researchers, including those at Amazon, discovered methods to 'jailbreak' the models' security features. Anthropic complied by shutting down access to all users to ensure compliance with the order.
Deeper Dive & Context
National Security Concerns
The Trump administration's decision was reportedly influenced by suspicions that a China-linked group may have accessed the Mythos 5 model. White House AI czar David Sacks stated that Anthropic prioritized model availability over addressing the security vulnerabilities, a claim the company disputes. Anthropic argued that the discovery of a narrow potential jailbreak should not justify recalling a widely deployed commercial model.
Industry and Expert Reactions
Amazon CEO Andy Jassy and other tech leaders had previously raised concerns about the security risks of Anthropic's models. Some experts criticized the sweeping nature of the restrictions, noting that it affects allies like Canada and the UK as well as adversaries. The order has raised constitutional and commercial concerns, particularly regarding the precedent of restricting software access.
Historical Context
The Trump and Biden administrations have previously imposed export controls on other technologies, such as semiconductors and supercomputers. However, this marks the first time the U.S. government has taken such sweeping measures to restrict access to AI models developed by an American company.
Anthropic's Response
Anthropic stated that the U.S. government's order was based on the discovery of a method to bypass the guardrails of Fable 5. The company warned that applying such a standard across the industry could halt all new model deployments. Researchers at Amazon had conducted jailbreak research that revealed vulnerabilities in the model, according to reports.
Long-Term Implications
The order has sparked debate over the balance between national security and the free flow of advanced technology. Some argue that the restrictions could stifle innovation and collaboration, while others support the measures as necessary to prevent misuse of powerful AI models.