The U.S. government has restricted foreign nationals' access to Anthropic's artificial intelligence models "Fable5" and "Mythos5", and South Korea's government has begun checking the facts.

On June 14, the Ministry of Science and ICT said it is determining specific details related to the restrictions on access to Fable5 and Mythos5. A ministry official said it is in ongoing communication with Anthropic and is confirming the facts.

Earlier, the U.S. government issued export control guidance restricting access by all foreign nationals to Anthropic's top-tier AI models, in line with national security authorities' guidelines. The measure is known to include not only access from overseas but also use of the models by foreign nationals within the United States. Some analysts also suggest the measure was influenced by Amazon CEO Andy Jassy informing government officials, including U.S. Treasury Secretary Scott Bessent, about vulnerabilities in Anthropic models.

Anthropic immediately suspended Fable5 and Mythos5 services for all customers to comply with the rules. Anthropic said the measure appears to stem from a misunderstanding of technical facts, and it plans to consult the U.S. government to restore the service.

The measure comes as the South Korean government has joined "Glasswing", an AI cooperation initiative led by Anthropic. Project Glasswing is a global AI security consultative body launched alongside the public release of Mythos. It operates by proactively providing models to vetted companies and institutions to defend against software vulnerabilities. In South Korea, Samsung Electronics, SK Hynix, SK Telecom and the Korea Internet & Security Agency (KISA) have joined Glasswing.

The ministry is expected to confirm the specific background and scope of the measure, its impact on South Korea and review future response directions.