Jade Leung

Jade Leung has helped build a technical function governments rarely possessed: the ability to test advanced AI systems independently of their developers. Her dual role links model evaluation to the centre of government, while keeping the distinction between scientific evidence and political decision-making visible.

Jade Leung works at the point where frontier-model testing becomes state capacity. As Chief Technology Officer of the United Kingdom’s AI Security Institute, she leads the technical organisation responsible for evaluating advanced systems across cyber capability, chemical and biological risk, safeguards, autonomy and societal effects. In August 2025, she was also appointed the Prime Minister’s AI Adviser, reporting directly to the Prime Minister and the Secretary of State for Science, Innovation and Technology while retaining her institute role. The arrangement gives technical evidence a short route into government, but it also makes clarity about institutional independence and attribution essential. Leung was one of the many named authors of the institute’s December 2025 Frontier AI Trends Report, its first public synthesis of results from two years of government testing. The underlying programme had evaluated more than 30 advanced models and reported rapid capability growth across several domains. Under her technical leadership, the institute expanded an open evaluation stack. By June 2026, its Inspect ecosystem offered more than 200 pre-built evaluations and had attracted over 240 contributors. The achievement is institutional: building people, tools and procedures through which government can produce its own evidence.