Today, the United Kingdom’s Frontier AI Taskforce announced a strategic partnership with OpenMined to develop and deploy technical infrastructure that will facilitate AI safety research across governments and AI research organisations.
A Growing Network of Technical Partnerships
According to the Taskforce’s second progress report, this collaboration is part of a broader initiative to partner with leading technical organizations in the AI field. The Taskforce has now established partnerships with 11 organizations, collectively building an extensive network of expertise in AI safety.
OpenMined joins other notable partners including ARC Evals, Advai, The Centre for AI Safety, Collective Intelligence Project, Faculty, Gryphon Scientific, RAND, Redwood Research, Trail of Bits, and Apollo Research. These partnerships specifically target organizations with expertise in understanding AI risks across various domains, including biosecurity, cybersecurity, and deceptive behavior.
Building State Capacity in AI Safety
The Taskforce, described as “a start-up inside government,” has been making rapid progress in building state capacity to evaluate risks at the frontier of AI. In just 18 weeks of operation, the team expanded from a single frontier AI researcher to a collective 150 years of frontier AI research experience across the team.
This partnership with OpenMined aligns with the Taskforce’s mission to address the challenges posed by increasingly capable AI systems. As the report notes: “We believe it is critical that frontier AI systems are developed safely and that the potential risks of new models are rigorously and independently assessed for harmful capabilities before and after they are deployed.“
Looking Forward: The AI Safety Institute
The progress report also announced that the UK government is putting its work on AI safety on a longer-term basis through the creation of an AI Safety Institute (AISI). This institute, described as “the first state-backed organisation focused on frontier AI safety for the public interest,” will continue the work of the Taskforce.
The institute’s mission is to “minimise surprise to the UK and humanity from rapid and unexpected advances in AI,” by developing the sociotechnical infrastructure needed to understand the risks of advanced AI and support its governance.
The partnership with OpenMined represents a significant step in this direction, focusing on building the technical tools and frameworks necessary for responsible AI governance in an increasingly complex technological landscape.
