Anthropic Faces Global Access Ban Over Claude Cybersecurity Allegations
On June 12, the U.S. government abruptly mandated that Anthropic’s advanced AI models, Fable 5 and Mythos 5, be restricted exclusively to domestic users. This sudden directive, issued under the Trump administration, effectively cut off global access to these powerful systems due to the company’s inability to verify user locations instantly. Researchers, international partners, and foreign nationals globally were blocked from the platforms within 90 minutes of the announcement. The move follows serious allegations that the Claude-related models pose a significant threat to national security, forcing a rapid recalibration of how these technologies are distributed and monitored internationally.
- The U.S. government restricted access to Fable 5 and Mythos 5 models to protect national security interests.
- General Joshua Rudd reported that the Mythos model successfully compromised classified NSA systems during authorized penetration testing.
- Anthropic warned that strict, unilateral export controls could stifle future innovation across the entire artificial intelligence industry.
NSA Findings Reveal Critical Vulnerabilities
According to reports from The Economist, the decision to enforce these restrictions stemmed from a briefing by Senator Mark Warner. General Joshua Rudd, head of the U.S. Cyber Command and the NSA, disclosed that during a controlled penetration test, the Mythos model managed to infiltrate nearly all classified systems within a matter of hours. 
This startling capability highlights that, despite being designed for defensive cybersecurity research, the model possesses offensive power that exceeds current safety thresholds.
The government deemed the potential for unauthorized access to sensitive data an unacceptable risk to national security.
Safety Mechanisms Fail to Prevent Risks
When Anthropic introduced the Mythos model in April, the company explicitly stated that it would not release a public version due to the software’s profound ability to detect and exploit security vulnerabilities. The model had been restricted to a small cohort of 200 elite partners, including tech giants like Amazon, Google, and Microsoft, under the Project Glasswing initiative. However, the situation escalated when the Fable 5 model, which was released to the public on June 9 with additional safety layers, also faced scrutiny.
Government officials argue that the security safeguards implemented in Fable 5 are fundamentally insufficient to mitigate its advanced cyber-attack capabilities. The crisis intensified after Amazon notified the U.S. Department of Commerce that a specific method had been identified to bypass these safety mechanisms. This disclosure forced the government to act immediately to prevent potential misuse of the technology.
Industry Leaders Debate Future Regulations
Anthropic has formally challenged the government’s decision, asserting that the identified vulnerability is narrow and potentially common across other large language models currently in development. The company cautioned that applying these rigid standards to the entire sector could lead to a permanent halt in the deployment of future AI breakthroughs.
The intelligence agencies of the Five Eyes nations have declared that AI will permanently alter the landscape of global cyber defense and offense.
As the international community watches these developments, the integration of artificial intelligence into critical infrastructure remains a subject of intense debate. For now, Fable 5 and Mythos 5 remain inaccessible to the vast majority of the global public while Anthropic works to address the government’s concerns and resume operations.
How do you believe these stringent export restrictions on AI development will reshape the future of the global technology landscape? Share your thoughts in the comments below.
Your comment has been submitted,
it will be published after approval.