Leading AI Developers and Governments Agree to Assess Risks Before Releasing New Models in Landmark Deal at UK Summit

Leading AI developers agreed to work with governments to test new models before release to manage risks. This was seen as a landmark achievement at the UK's AI summit.

Political leaders from the US, EU, and China agreed to share an approach to identifying AI risks and mitigating them.

Yoshua Bengio will help deliver a report on AI capabilities and risks to build a shared understanding.

Companies agreed with governments that models should be rigorously assessed before and after deployment.

China was represented at the summit but did not sign Thursday's testing agreement.