UK Establishes World's First AI Safety Institute to Evaluate Frontier Models
The UK launches a global AI Safety Institute to rigorously test emerging AI models for potential harms. Leading tech companies and international partners back the new permanent research hub.
The UK officially launches the world's first AI Safety Institute to evaluate the risks of emerging AI models. This new global hub evolves from the Frontier AI Taskforce and operates under the continued leadership of Chair Ian Hogarth. An external advisory board of experts in national security and computer science guides the institute's ongoing mission.
The institute carefully tests new frontier AI models both before and after public release to identify potentially harmful capabilities. Researchers investigate a wide range of risks, covering social harms like bias and misinformation to extreme threats such as humanity losing control of AI completely. The organization partners closely with the Alan Turing Institute to conduct this advanced data science research.
Major AI companies and international nations, including the US, Singapore, and Google DeepMind, agree to collaborate with the new institute. This global partnership cements the UK's position as a world leader in AI safety and helps develop the most advanced AI protections of any country. The government states this initiative ensures the British people can safely capture the countless benefits of AI for future generations.