US Government Orders Anthropic to Halt Access to Powerful AI Models
AI firm Anthropic has disabled access to its Claude Fable 5 and Claude Mythos 5 models globally following a US government directive citing national security concerns. The order mandates that all foreign nationals, including employees, cease using these models, leading to a complete shutdown for all users to ensure compliance. Anthropic believes the action stems from concerns over a potential 'jailbreaking' method.
Key points
- AI company Anthropic has suspended access to its Claude Fable 5 and Claude Mythos 5 models worldwide.
- The action follows an export control directive from the US government citing national security concerns.
- The directive required suspending access for all foreign nationals, leading Anthropic to disable the models for all users to comply.
- Anthropic believes the government's concern relates to a discovered method for bypassing Fable 5's security safeguards.
- Mythos 5 was a restricted model for cybersecurity research, while Fable 5 was a more broadly released version with safety guardrails.
AI developer Anthropic has complied with a US government order to immediately disable access to its Claude Fable 5 and Claude Mythos 5 models. The directive, received Friday at 5:21 PM ET, cited national security concerns and mandated that all foreign nationals, including those within Anthropic, cease using the advanced AI systems. To ensure compliance, Anthropic stated it had to suspend access for all its customers globally.
The company expressed disagreement with the directive, noting that the government's letter did not specify its national security concerns. Anthropic's understanding is that the government believes a method has been found to bypass or "jailbreak" Fable 5, potentially aiding in hacking. Anthropic, however, reviewed a demonstration of such a technique and found it identified only previously known, minor vulnerabilities, comparable to what other publicly available models could discover.
Mythos 5 is Anthropic's most advanced AI, developed for defensive cybersecurity work and shared with approximately 50 vetted organizations through its Project Glasswing initiative. Fable 5, launched just three days prior, was presented as a more commercially accessible version of Mythos, incorporating robust guardrails intended to prevent misuse in high-risk areas like cybersecurity. Despite these safeguards, which Anthropic noted some users found overly restrictive, the government has ordered their suspension.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.