Anthropic said it disabled access to its frontier models Fable 5 and Mythos 5 after receiving a U.S. export-control directive citing national security authorities, the company wrote in a June 12 blog post. The directive, which Anthropic said it received at 5:21 p.m. ET, ordered the company to suspend access to those models for any foreign national, including foreign-national employees; Anthropic wrote that to comply it had to turn off the models for all customers. Anthropic said the government did not provide detailed technical findings but told the company it had become aware of a method to bypass safeguards. Reporting from CNBC, Bloomberg, Fortune and NBC corroborates the company statement and the Commerce Department role in issuing the order.
What happened
Anthropic wrote in a blog post published Jun 12, 2026 that the US government issued an export-control directive instructing the company to suspend access to Fable 5 and Mythos 5 for any foreign national. Anthropic said it received the directive at 5:21 p.m. ET and, because the order covered foreign nationals whether inside or outside the United States, the company disabled access to both models for all customers to ensure compliance. Anthropic wrote that access to its other Claude models would not be affected. (Anthropic blog post; CNBC; Fortune)
Anthropic stated the government did not provide specific technical details in the directive. The company wrote that government officials said the decision followed a demonstration of a method that could bypass Fable 5 safeguards; Anthropic said that demonstration identified a small number of previously known, relatively simple vulnerabilities and that similar behavior can be elicited from other publicly available models. (Anthropic blog post; Fortune)
Reporting by NBC and Bloomberg described involvement by the Commerce Department in issuing the order. NBC reported that the letter issuing the directive was sent from the Commerce Department, citing an administration official. (NBC; Bloomberg)
Editorial analysis - technical context
Frontier foundation models combine increased capabilities with safety guardrails that are continuously tested by red teams and external researchers. Publicly reported incidents where researchers or adversaries demonstrate targeted jailbreak techniques tend to be narrow in scope yet attract outsized scrutiny because they expose high-risk capability vectors, especially for tasks related to cybersecurity. Industry participants routinely weigh the trade-off between capability and restrictive safety layers; forced, external restrictions from regulators change that risk calculus for all providers. (Industry-pattern observations)
Context and significance
Government use of export controls and national-security authorities to limit access to advanced AI models represents a regulatory escalation. Observers will compare this action to prior controls on dual-use technologies and to recent policy proposals that aim to treat some frontier models as strategically sensitive. For practitioners, the episode underscores the likelihood that model availability and compliance regimes may become an operational constraint for cloud deployments, partnerships, and international research collaborations. (Industry context)
What to watch
- •Whether the Commerce Department or another agency publishes technical findings or an explanatory memo that clarifies the scope and nature of the alleged bypass.
- •If other frontier providers receive similar directives or if vendors proactively restrict deployments in jurisdictions subject to export controls.
- •Follow-up technical writeups or remediation reports from Anthropic, external red teams, or independent researchers that demonstrate whether the reported bypass generalizes beyond the narrow case Anthropic described.
- •Changes to customer SLAs, partner onboarding, and enterprise risk assessments for models that may be subject to export-control restrictions.
Bottom line
This is an unusually direct instance of national-security authorities intersecting with commercial model deployments. The immediate operational outcome is a shutdown of two of Anthropic's leading models; the broader effect is to put regulatory risk squarely into deployment and launch planning for frontier models across the industry.
Scoring Rationale #
This is a high-impact regulatory intervention affecting frontier-model availability. It directly affects deployments and signals heightened government willingness to use export controls on AI, which matters for practitioners and vendors.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.