UK AI Safety Institute Gains OpenAI and Microsoft as Partners
The UK's AI Safety Institute (AISI) has expanded its global coalition, with OpenAI and Microsoft joining and pledging financial support for its Alignment Project. The institute's alignment fund has now reached £27 million to support 60 research projects focused on ensuring AI systems are controllable. This follows a recent AI Safety Summit where industry leaders convened to discuss cross-border cooperation on risk frameworks.
- The recent funding injection from OpenAI, valued at £5.6 million, and additional support from Microsoft were announced at the AI Impact Summit in New Delhi. This brings the total funding for the Alignment Project to over £27 million. - The Alignment Project is an international research initiative that also includes backing from Amazon Web Services (AWS), Anthropic, the Canadian AI Safety Institute, and Schmidt Sciences. - The project's core research questions focus on two key challenges: how to prevent AI systems from carrying out harmful actions and how to design systems that do not attempt such actions in the first place. - The November 2023 AI Safety Summit, held at Bletchley Park, resulted in the "Bletchley Declaration," a commitment by 28 countries and the EU to cooperate on identifying and mitigating frontier AI risks. - As a direct outcome of the summit, leading AI companies agreed to grant governments early access to their models for safety evaluations, a significant step toward transparency. - The UK's AI Safety Institute (AISI) was the first state-backed organization focused on advanced AI safety and aims to develop the infrastructure needed to understand and govern the risks of advanced AI. - The AISI has since been followed by the creation of similar institutes in the US, Japan, Canada, and other nations, forming an international network to advance the science of AI safety. - In early 2025, the UK's institute was rebranded as the AI Security Institute (AISI) to reflect a heightened focus on national security applications and potential threats, a move that followed similar shifts by major AI companies.
