UC Berkeley's multidisciplinary research center advancing AI safety, agentic AI, and decentralization technology to empower a responsible digital economy.
UC Berkeley's multidisciplinary research center advancing AI safety, agentic AI, and decentralization technology to empower a responsible digital economy.
People
Updated 05/18/26Co-Director
Co-Director
Funding Details
Updated 05/18/26- Annual Budget
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26The Berkeley Center for Responsible, Decentralized Intelligence (RDI) is a multidisciplinary, campus-wide initiative at the University of California, Berkeley, established around 2021. It is co-directed by Professor Dawn Song (EECS, known for her work in AI safety, security, and decentralization) and Professor Christine Parlour (Haas School of Business, finance). The center draws on faculty from UC Berkeley's College of Engineering, Haas School of Business, and School of Law, as well as affiliate faculty from institutions including Imperial College London, ETH Zurich, UIUC, and NYU. RDI is organized around three pillars. The research pillar covers agentic AI foundations (reasoning, planning, reinforcement learning, self-improvement), AI safety and security (robustness, adversarial resilience, oversight, alignment, control, interpretability, monitoring), agentic AI applications (web agents, coding agents, multi-agent systems), AI for cybersecurity, program synthesis, and decentralization technology (cryptography, DeFi, privacy, zero-knowledge proofs). The education pillar includes popular MOOCs on LLM agents and agentic AI that have enrolled over 40,000 learners globally, as well as a Zero Knowledge Proofs MOOC and a certificate program in Decentralization, AI and Web3. The community and entrepreneurship pillar is anchored by the Berkeley Xcelerator, which has incubated over 110 teams with more than $650 million in follow-on funding secured. RDI has received significant external funding, including a $5 million grant from the Algorand Foundation in 2022 to establish the Berkeley RDI-Imperial College London Algorand Center of Excellence focused on scalability, privacy, and responsible decentralized finance. Annual summits on decentralization and AI and agentic AI have drawn thousands of attendees from industry and academia. The center's Discord community has grown to approximately 14,000 members, and its hackathons and competitions have attracted over 1,300 teams from more than 100 countries with prize pools exceeding $1.5 million. As a UC Berkeley research center, RDI does not operate as an independent nonprofit; it is funded through university channels, grants, and industry partnerships.
Theory of Change
Updated 05/18/26RDI believes that making AI systems safe, robust, and beneficial requires both technical research and broad education. By conducting foundational research on AI safety, alignment, interpretability, and adversarial robustness at a leading research university, RDI aims to produce knowledge and tools that can be adopted across the AI field. Through massive open online education, the center seeks to train a large global cohort of practitioners who understand responsible AI development. Through its Xcelerator and summit programs, it aims to translate research into companies and policies that embed safety-conscious approaches from inception. The decentralization angle adds a structural mechanism: distributing AI development and data ownership reduces single points of failure and concentrations of power that could otherwise lead to unsafe or misaligned outcomes.
Grants Received
Updated 05/18/26Projects– no linked projects
Updated 05/18/26Discussion
No comments yet. Be the first to share your thoughts.