US Government Restricts Access to Anthropic AI Models Over Security Concerns

The U.S. government has mandated that AI company Anthropic restrict access to its latest models, Fable 5 and Mythos 5, citing potential national security risks. The directive, issued in mid-June, requires the company to block these models from being accessed by foreign nationals, a measure that forced Anthropic to temporarily disable the services for all users due to technical limitations in verifying user nationality in real-time. This intervention marks a significant escalation in the ongoing regulatory scrutiny of high-performance artificial intelligence.

Why Did the Government Intervene?

The federal government’s action stems from concerns regarding potential “jailbreaks” that could allow users to bypass safety guardrails. According to reports from Semafor and Fortune, researchers at Amazon—a major investor in and infrastructure provider for Anthropic—demonstrated that the models could be manipulated to disclose restricted information related to cyberattacks. Following these findings, Amazon CEO Andy Jassy reportedly raised the issue directly with the administration. While Anthropic has characterized the vulnerability as “narrow and not universal,” the government maintains that the risk necessitates a pause in deployment until the security gaps are fully addressed.

How Have Export Controls Affected Operations?

The government issued a formal export control order prohibiting “any foreign national” from accessing Fable 5 and Mythos 5, regardless of whether they are located within or outside the United States. Because current API infrastructure does not verify the citizenship of every individual user in real-time, Anthropic opted to disable the models entirely to ensure compliance. This move has impacted domestic users, who were originally intended to maintain access. The company is currently working to reinstate availability for authorized users while navigating the regulatory requirements set by the Department of Commerce.

The Regulatory Context: A Strained Relationship

This mandate follows months of friction between Anthropic and the Trump administration. The conflict centers on disagreements over the use of AI in national security contexts. Anthropic has historically maintained strict policies against the use of its technology in autonomous weapons systems and mass surveillance. Conversely, the Pentagon has pushed for broader access to AI capabilities for “all lawful purposes.” In previous months, the administration labeled Anthropic a “supply-chain risk,” a designation that led to legal challenges in federal court. While a district court judge initially sided with Anthropic, an appellate court later allowed the government’s restrictions to proceed.

What Are the Broader Industry Implications?

The restriction of Fable 5 and Mythos 5 has sparked pushback from industry experts. A group of security professionals from organizations including Nvidia, Zoom, and Mercedes-Benz recently published an open letter advocating for a more transparent and scientific approach to AI risk assessment. Critics argue that limiting access to advanced models while global competitors continue to develop their own capabilities could disadvantage U.S. firms. As the administration signals that future models exceeding specific performance thresholds may undergo mandatory government review, the industry faces a shift toward increased federal oversight of AI development lifecycles.

Key Developments

• Regulatory Trigger: A documented “jailbreak” allowed researchers to bypass safety filters to access sensitive cybersecurity information.
• Scope of Restriction: Access to Fable 5 and Mythos 5 was suspended globally to comply with export controls regarding foreign national usage.
• Legal Precedent: The action follows a series of disputes between Anthropic and the Pentagon regarding the integration of AI into military and surveillance operations.
• Industry Response: Security experts are calling for a standardized, transparent process for evaluating AI risks to avoid blunt-force service shutdowns.

The situation remains fluid as Anthropic continues discussions with the Department of Commerce. Whether the models will be restored remains dependent on the company’s ability to satisfy government security benchmarks and implement technical controls that satisfy federal requirements.