Show summary Hide summary
Anthropic took two of its newest artificial intelligence models offline on Friday after receiving a U.S. government order intended to stop their use by non‑U.S. nationals. The move marks a sharp escalation in Washington’s effort to limit international access to the most advanced AI systems—and has immediate consequences for researchers, customers and competitors.
What happened
Pittsburgh fireworks guide: find the best viewing sites and legal ignition zones for July 4
Delivery impersonator shoots through package at ex’s new boyfriend
The company disabled access to Fable 5 and Mythos 5 after being told by federal officials to prevent their use by foreign users. Fable had been released broadly earlier this week; Mythos remained tightly restricted by Anthropic for security and cybersecurity concerns.
Anthropic said the directive arrived Friday afternoon and did not identify the specific national‑security issues behind the order. The company criticized the way the government handled the matter, arguing that any intervention to block risky deployments should follow a clear, fair and technically grounded legal process.
The Commerce Department did not immediately reply to requests for comment.
Why this matters now
Washington’s instruction comes just 10 days after the president signed an executive order creating a voluntary review process for the most capable AI models. That framework lets federal officials examine advanced systems for up to a month before a developer can release them publicly—although companies are not required to participate.
The Anthropic action is the first clear instance of export controls being applied to specific AI models, and it signals how quickly policy decisions can affect product availability and international research collaboration.
| Model | Availability before directive | Access after directive | Company posture |
|---|---|---|---|
| Fable 5 | Widely released to users | Taken offline for foreign users | Disagrees with government’s handling; seeking clarity |
| Mythos 5 | Already tightly restricted due to cybersecurity concerns | Access further curtailed | Access previously limited; now paused for compliance |
Immediate implications
Developers, academics and companies that were testing or integrating these models face abrupt interruptions. The move also raises questions about how U.S. export rules will be applied to AI and which actors—commercial or academic—might be affected in future enforcement actions.
- Research slowdown: International teams may lose access to tools they were using for experimentation and evaluation.
- Commercial disruption: Partners relying on model access for products or services could see short‑term outages or contract complications.
- Policy precedent: This is the most concrete step yet toward restricting cutting‑edge AI abroad, potentially shaping how other governments react.
What Anthropic says and next steps
Anthropic has pushed back publicly, saying the government’s directive lacked transparency and did not outline the national‑security basis for the restriction. The company described the situation as a misunderstanding and indicated it is working to restore service as quickly as possible.
Beyond the immediate dispute, the episode highlights a broader tension: firms want predictable rules that balance safety with innovation, while regulators are grappling with how to act quickly in a fast‑moving technology landscape. How the administration applies its export controls going forward—and whether companies will seek judicial or legislative remedies—will determine whether this becomes an isolated incident or the start of a new regulatory norm.
Officials have a short window to clarify expectations and procedures if they intend to use export controls regularly; without that clarity, companies and international partners may face further uncertainty.
