The UK government's AI Safety Institute is expanding internationally by opening its first overseas office in San Francisco this summer.

The new office will allow the Institute to tap into the technology talent in the Bay Area, engage with major AI labs headquartered in London and San Francisco, and cement relationships with the United States to advance AI safety for the public interest.

The San Francisco office, expected to open in the coming months, will recruit an initial team of technical staff led by a Research Director. It will complement the Institute's London headquarters, which continues to grow and now has over 30 employees focused on assessing the risks of cutting-edge AI systems.

By establishing a presence in the US, the Institute aims to foster close collaboration, further the strategic partnership between the two countries on AI safety, share research, and conduct joint evaluations of AI models to inform global AI safety policies.

Announcing the expansion, Science and Technology secretary Michelle Donelan said, "This expansion represents British leadership in AI in action. It is a pivotal moment in the UK's ability to study both the risks and potential of AI from a global lens, strengthening our partnership with the US and paving the way for other countries to tap into our expertise as we continue to lead the world on AI safety."

The Institute has also released selected results from recent safety testing of five publicly available advanced AI models, making it the first government-backed organisation to unveil such evaluations. The tests assessed models across key risk areas, including the effectiveness of installed safeguards.

Findings included models completing some cybersecurity challenges but struggling with advanced ones, demonstrating PhD-level knowledge in chemistry and biology, vulnerability to basic "jailbreaks" enabling harmful outputs, and inability to complete complex, lengthy tasks without human oversight.

Institute Chair Ian Hogarth said the results contribute to assessing model capabilities and lack of robustness in existing safeguards, adding, "Our ambition is to continue pushing the frontier of this field by developing state-of-the-art evaluations, with an emphasis on national security related risks."

---