# US Imposes Export Controls on Frontier AI Models

Over the next 3-6 months Washington extends export-control authority to specific frontier AI systems, forcing US developers to restrict foreign-national access and treating advanced models as controlled strategic technology.

- Conviction: 10 / 100 (weakening)
- 7-day move: -14
- 30-day move: -58
- Horizon: Medium term (3-9 months)
- Tracking since: 2026-06-13T00:00:00.000Z
- Last updated: 2026-07-22T05:50:04.606Z
- Canonical: https://polylog.news/trends/us-ai-model-export-controls
- Publisher: Polylog
- Status: invalidated (No move to place frontier AI models under export control has materialized in 7+ days, and today's news shows capital and model development flowing to non-US labs (Mistral, Moonshot), undercutting the premise that Washington can gate frontier AI as controlled strategic technology.)
- Affected regions: United States

## Recent score history

- 2026-07-20: 12
- 2026-07-21: 10

## Recent evidence

- [contradicts] OpenAI and Google Supplied AI Models to Blacklisted Chinese Groups, FT Reports (2026-07-10): The FT reports OpenAI and Google sold AI services to Singapore-based subsidiaries of restricted firms Alibaba, Baidu and Tencent—showing frontier-model controls are porous and being routed around, weakening the thesis that Washington is effectively treating models as controlled strategic tech.
- [confirms] The AI Trade Faces a Reckoning as Bubble Fears Grow and Washington Tightens Its Grip (2026-06-27): Washington permitted only a restricted release of Anthropic's most powerful model, continuing to treat frontier AI systems as controlled strategic technology.

2 more evidence entries, the full score history, the conviction-driver timeline, and affected assets are for subscribers: https://polylog.news/pricing