Summary
The UK announced the first government AI Safety Institute, tasked with testing frontier AI models for dangerous capabilities before public release. The institute worked directly with AI companies — including OpenAI and Anthropic — and shared findings with allied governments.
The institute's mandate, its relationship with AI companies, and how it differs from a regulatory body.
Detailed account of what the institute actually does and what it has found in its model evaluations.
Chip war drives AI governance race
US–China Strategic Competition