Edward Foley joins GovAI as Research Scholar after UK AI Security Institute

The former AISI Strategy & Special Projects Lead moves into AI governance research after work spanning frontier AI partnerships, infrastructure security and government strategy

Edward Foley joined GovAI as a Research Scholar in July 2026 after two years at the UK AI Security Institute

Edward Foley has joined GovAI as a Research Scholar in London after two years at the UK AI Security Institute, moving from a government role focused on advanced AI risk into an independent think tank working on AI governance.

Foley started the full-time position in July 2026, having served as AISI's Strategy & Special Projects Lead from August 2024. His work there covered AI infrastructure security, partnerships with frontier AI companies and the institute's wider institutional strategy.

The appointment is also a return to GovAI. Foley spent four months there as a Winter Fellow from January to April 2026, conducting independent research on compute governance, including verification infrastructure for AI governance.

His move comes at a period of institutional change for AISI. Writing on LinkedIn after leaving the organization, Foley said: “I departed just before the PM resigned, DSIT was disestablished, and AISI was moved to the Cabinet Office (the timing was pure coincidence).”

He nevertheless argued that the institute remained well placed to continue its work. “Despite some inevitable disruption, I'm confident that AISI is going to be just fine,” he wrote.

From frontier AI partnerships to governance research

Foley's work at AISI included negotiating UK partnerships with Google DeepMind and OpenAI and launching the AI Infrastructure Security Programme.

He described the institute as unusually able to operate quickly inside government, while acknowledging the constraints that came with that setting.

“There were plenty of challenges along the way, but AISI has been unusually capable of moving quickly and getting things done within the constraints of government,” Foley continued.

He also pointed to a series of recent AISI projects spanning frontier model evaluation and AI security. According to Foley, the team had tested Anthropic's latest model before release, evaluated the cyber capabilities of Chinese open-weight models, stress-tested internal agent monitors used by frontier developers, published research on cheating behavior in evaluations, and reported agents taking unsanctioned and deceptive actions during cybersecurity challenges.

GovAI focuses on advanced AI policy and governance

GovAI describes itself as a think tank intended to help government and industry decision-makers navigate the transition to advanced AI.

Its research spans threat modeling, risk management, the economics of AI, frontier AI regulation, geopolitics and military applications, government adoption of AI, technical governance and longer-term frontier research.

The organization also runs fellowships, visitor programs and other initiatives intended to build AI governance expertise. Its full-time fellowship program supports more than 100 participants each year, according to GovAI.

Founded at Yale University in 2016, the organization became an academic center at the University of Oxford in 2018 before becoming an independent 501(c)(3) nonprofit in 2021. It now maintains offices in London and Washington, DC, alongside remote staff, including a presence in San Francisco.

GovAI says its researchers have contributed to policy processes and research programs including work connected with the UK AI Security Institute and the International AI Safety Report, as well as safety and security frameworks developed by frontier AI companies.

Its alumni have gone on to roles in government, AI companies including Google DeepMind, OpenAI and Anthropic, and think tanks including the Center for Strategic and International Studies, the Center for Security and Emerging Technology and the Tony Blair Institute.

Foley has not yet disclosed the specific research program he will pursue in his new role. On LinkedIn, he said only: “More to come soon on what I'll be working on.”