U.S. Export Controls Shaped AI Model Access Policies
New government directives led OpenAI and Anthropic to restrict model access to address potential security vulnerabilities.
Updated on Sept. 24, 2026 in Artificial Intelligence
Live Poll
Should the U.S. prioritize strict national AI controls over international safety testing collaboration?
U.S. government export control directives issued on June 12, 2026, forced Anthropic to suspend access to its Claude Fable 5 and Mythos 5 models for foreign nationals. These measures aimed to mitigate potential jailbreak vulnerabilities in advanced artificial intelligence systems.
Why it matters
The restrictions reflect increasing regulatory scrutiny over high-level AI deployment to prevent the misuse of powerful models. By requiring companies like OpenAI to share participant identities, the U.S. government aims to ensure rigorous pre-deployment evaluation.
Anthropic employs Fable and Mythos safeguards to manage model behavior, with Fable testing spanning thousands of hours. OpenAI has restricted GPT-5.6 Sol access to trusted partners and the UK AI Security Institute for pre-deployment evaluation.
The players
Anthropic
This American AI safety and research company develops large language models and is a significant competitor to OpenAI in the generative AI space.
OpenAI
This artificial intelligence research organization is responsible for creating the GPT series of models and leading the current industry transition toward advanced large language models.
UK AI Security Institute
This government-backed entity is tasked with evaluating the safety and security of advanced AI models before they are deployed publicly.
The details
Anthropic now limits Mythos 5.1 model access to select U.S. organizations to comply with U.S. export controls. OpenAI concurrently initiated a limited preview for its GPT-5.6 Sol model on June 26, 2026, granting specific access to the UK AI Security Institute.
Timeline
Anthropic suspended foreign national access on June 12, 2026.
OpenAI initiated the GPT-5.6 Sol preview on June 26, 2026.
A UK parliamentary question regarding Anthropic was submitted on Sept. 16, 2026.
The UK government deadline to answer the parliamentary question is Sept. 30, 2026.
The Tech Race
These restrictions highlight a broader shift where AI development is increasingly subjected to the same export control frameworks traditionally reserved for physical hardware. This marks a departure from the previously unregulated environment, placing AI safety protocols at the center of international trade and national security strategy.
Users may experience limited access to the most advanced AI tools as companies implement stricter verification processes for international users. Researchers and organizations should anticipate longer pre-deployment evaluation timelines as safety testing becomes a regulatory requirement.
The takeaway
Advanced AI models are increasingly treated as strategic national assets subject to rigorous regulatory oversight. Organizations must be prepared to navigate complex international compliance requirements to continue utilizing cutting-edge intelligence technologies.
What happens next
The UK government is scheduled to provide a formal response to a parliamentary question regarding Anthropic's cooperation with safety testing by September 30, 2026.
Further reading
Learn more about the evolving landscape of global standards in Artificial Intelligence.
Live Poll
Should the U.S. prioritize strict national AI controls over international safety testing collaboration?
