Wire · regulatory
UK AI minister details how the restored frontier model testing deal works
◆ Sectors
AI Frontier Models
◆ Geography
United Kingdom
◆ Source
◆ Verified
Fusion42 · 12 October 2026 · Fusion42 review
The UK AI Security Institute’s pre-release access to new frontier AI models from OpenAI, Anthropic, and Google DeepMind has been restored after a 17-day US freeze, reverting to the voluntary testing terms that existed before the freeze.
This Wire brief sits within Fusion42's coverage of AI Frontier Models.
◆ ◆ The Wire takeaway
You gain restored access to early testing of top AI models from US labs, reopening a key route for UK AI safety evaluation that was briefly cut off. Use this window to strengthen your model audit and alignment capabilities ahead of any future regulatory tightening.
◆ Coverage
1 source · 12 Oct 2026
◆ Related on Wire
◆ Topics
AI Frontier Modelsukai-security-institutemodel-testingopenaianthropicgoogdeepmind