University of Oxford
Governance of AI (GovAI) Program
United Kingdom · DPhil/Postdoc/Fellowship
GovAI is the University of Oxford's flagship research program focused on AI governance, studying governance mechanisms for frontier AI systems, international coordination, and policy design. Led by Allan Dafoe, it is one of the most influential academic programs in AI governance, providing institutional safeguards and policy frameworks for AI alignment.
学术历史
GovAI was originally established in 2017 within the Future of Humanity Institute (FHI), founded by Allan Dafoe. FHI was founded by Nick Bostrom in 2005 and had long studied existential risks and superintelligence governance. The establishment of GovAI marked the transition of AI governance from philosophical discussion to systematic policy research. Following the closure of FHI in 2024, GovAI's research capacity was distributed across other Oxford centers (including the Institute for Ethics in AI and the Global Priorities Institute) and affiliated partner institutions. During its time at FHI, GovAI produced a substantial body of foundational AI governance literature, including agenda-setting papers on AI governance and research on international AI coordination mechanisms.
当前状态
GovAI's core research agenda includes: frontier AI model evaluation and audit frameworks, AI international governance mechanisms (AI agreements analogous to nuclear non-proliferation treaties), institutional design for AI regulation, and AI safety standards development. Affiliated researchers continue to conduct research across multiple Oxford centers. GovAI regularly publishes AI governance policy reports and maintains close collaboration with policy institutions including the UK AI Safety Institute and the EU AI Office. The program offers DPhil training schemes and Postdoc Fellowships.
实验室 / 研究中心
Governance of AI Program
访问页面 →关键人物
Allan Dafoe
Associate Professor / Founder of GovAI
Pioneer in AI governance and international relations, founder of the GovAI program. Researches AI international governance mechanisms, AI safety policy, and global coordination. Published multiple agenda-setting papers on AI governance. Formerly Head of Governance Research at DeepMind.
Nick Bostrom
Professor
Philosopher and existential risk research expert, founder of FHI. Author of "Superintelligence" (2014), a foundational work in AI safety and governance that defined the theoretical framework for the superintelligence alignment problem.
Miles Brundage
Researcher (former)
AI policy research expert, formerly conducted AI governance research at GovAI. Later served as Head of Policy Research at OpenAI, translating academic governance research into corporate practice.
标志性成果
- Published the agenda-setting AI governance paper (Dafoe et al., 2018), defining the core problems and directions of AI governance research
- Nick Bostrom's "Superintelligence" (2014): a foundational work in AI safety and governance
- Research on AI international governance mechanisms: proposing international coordination frameworks for AI analogous to arms control agreements
- Policy frameworks for frontier AI model evaluation and auditing, influencing the establishment of the UK AI Safety Institute
学术资源
证据
排名
QS World University Rankings #Top 2 globally in Politics and International Studies (2024)
出处：QS World University Rankings →师资
6 位相关教师
知名：Allan Dafoe、Nick Bostrom、Miles Brundage、Seán Ó hÉigeartaigh、Cullen O'Keefe
研究产出
GovAI-affiliated researchers continue to publish AI governance papers and policy reports in journals including AI & Society and Politics and AI, and at the AIES conference
就业去向
Graduates join the DeepMind governance team, OpenAI policy team, the UK AI Safety Institute, policy institutions, or take faculty positions in academia
发现信息有误？ 提交纠错