US restores UK AISI access to latest US models for safety testing

The US has restored the UK AI Safety Institute's access to the latest American frontier models for safety testing. A month earlier, Washington had ordered OpenAI, Anthropic and Google DeepMind to stop sharing models with UK regulators.

• US reopens AISI access to frontier models for safety testing
• Washington had barred OpenAI, Anthropic and Google DeepMind from sharing models
• The restriction lasted about a month