Harvard University

Harvard University, founded in 1636 in Cambridge, Massachusetts, is the oldest institution of higher learning in the United States and home to several significant programs in AI safety, AI governance, and related research. The Kempner Institute for the Study of Natural and Artificial Intelligence was established at Harvard following a $500 million, 15-year pledge from Mark Zuckerberg and Priscilla Chan through the Chan Zuckerberg Initiative, announced in December 2021. The institute is co-directed by computer scientist Sham Kakade and neuroscientist Bernardo Sabatini and is based at the Science and Engineering Complex in Allston. Its 2025 Annual Report documented 420+ community members, 350+ papers, 750,000+ compute jobs, and 3 million+ GPU hours. The institute organizes its research around three themes: Innovation in AI, Science of AI, and AI and the Brain, with the Kempner AI Cluster providing substantial compute infrastructure. The Berkman Klein Center for Internet & Society, founded in 1996, is a research center at Harvard Law School focused on internet policy, AI ethics, and digital governance. It received $5.9 million (shared with MIT's Media Lab) as an anchor institution for the Ethics and Governance of Artificial Intelligence Fund in 2017, a $27 million initiative backed by the Knight Foundation, Omidyar Network, Reid Hoffman, the Hewlett Foundation, and others. The Center's AI Initiative is directed by Jonathan Zittrain and conducts evidence-based research to guide policymakers and institutions in the responsible development and deployment of AI systems. The Center has welcomed over 500 fellows from 40+ countries and produces ongoing research on AI accountability, algorithmic fairness, and autonomous systems governance. The Harvard AI Safety Team (HAIST), later operating as the AI Safety Student Team (AISST) through the Cambridge Boston Alignment Initiative, supports undergraduate and graduate students conducting research to reduce risks from advanced AI. Open Philanthropy has provided multiple grants to support the team, including approximately $550,000 for education and outreach and $1.17 million for operating costs. The team runs introductory AI safety reading groups, a Governance, Policy and Strategy Fellowship, and technical research projects on anomaly detection and mechanistic interpretability. Harvard Kennedy School contributes through AI governance policy courses, executive education programs (Leading in Artificial Intelligence: Exploring Technology and Policy), and policy research through the Mossavar-Rahmani Center for Business and Government. SEAS faculty including Hima Lakkaraju (interpretability research) and Seth Neel (SAFR AI Lab, covering privacy, bias, reliability, and interpretability) conduct relevant technical safety research. Harvard also offers CS 2881 AI Safety, a graduate course covering mechanistic interpretability and AI safety research. Harvard Undergraduate Effective Altruism and related student groups further build a pipeline of students oriented toward high-impact AI safety careers.