University of California, Berkeley

The University of California, Berkeley is widely regarded as the world's leading public research university and serves as a major global hub for artificial intelligence safety research. Located in Berkeley, California, the campus hosts an ecosystem of interconnected research centers, labs, and initiatives dedicated to ensuring that AI development proceeds safely and in alignment with human values. The Center for Human-Compatible Artificial Intelligence (CHAI), founded in 2016 by Professor Stuart Russell along with co-principal investigators including Pieter Abbeel, Anca Dragan, Tom Griffiths, Bart Selman, Joseph Halpern, Michael Wellman, and Satinder Singh Baveja, is CHAI's flagship AI safety center. CHAI's mission is to develop the conceptual and technical foundations needed to reorient AI research toward provably beneficial systems. Its key research approach involves value alignment through inverse reinforcement learning — training AI systems to infer and respect human preferences from observed behavior, rather than assuming a fixed objective. CHAI has received over $17 million in funding from Open Philanthropy alone as of 2022, plus additional support from the Future of Life Institute, Survival and Flourishing Fund, the Leverhulme Trust, and CITRIS. Berkeley AI Research (BAIR) is the university's primary AI research laboratory, comprising more than 50 UC Berkeley faculty and over 300 graduate students working across computer vision, machine learning, natural language processing, planning, control, and robotics. BAIR also hosts the Responsible AI (Re-AI) Initiative, which advances understanding of transparency, fairness, privacy, safety, security, and accountability in AI systems. The Center for Long-Term Cybersecurity (CLTC) houses the AI Security Initiative, founded in 2019, which functions as a multidisciplinary research center focused on AI risk management standards and policy. CLTC also runs the AI Policy Hub in collaboration with the CITRIS Policy Lab, training Berkeley researchers to develop governance frameworks for AI. Berkeley RDI (Center for Responsible, Decentralized Intelligence) focuses on AI and agentic AI education and research, running large-scale open online courses (including one with 40,000+ registered learners) and an incubation program that has supported over 110 global teams. Berkeley is also home to the Berkeley Existential Risk Initiative (BERI), which provides operational support to university research groups working to reduce existential risks. Across these programs, UC Berkeley's total AI safety research community involves hundreds of faculty, researchers, and graduate students, making it one of the most productive and influential nodes in the global AI safety ecosystem.