University of Southern California

The University of Southern California (USC) is one of the oldest and most prominent private research universities in California, founded in 1880 in Los Angeles. With approximately 46,000 students, 4,600 full-time faculty, and 23 schools and academic divisions, USC is a major research institution with an endowment exceeding $8 billion and nearly $1 billion per year in sponsored research. USC's relevance to AI safety and existential risk reduction centers on several research groups and initiatives. The university received $512,000 through the Survival and Flourishing Fund's 2024 Flexible Hardware-Enabled Guarantees (flexHEGs) round, funded by the Future of Life Institute ($493,000) and Blake Borgeson ($19,000). This grant supported research on hardware-enabled governance mechanisms for AI accelerators used in large-scale AI training, which are designed to enable multilateral, privacy-preserving, and trustworthy verification and automated compliance guarantees for agreements regarding the development and use of advanced AI technology. USC hosts multiple research labs with direct relevance to AI safety. The SCIP Lab, led by Professor Murali Annavaram, conducts research in hardware security, machine learning systems, computer architecture, and private machine learning, including work on training ML models with private data on untrusted hardware. The Safe Autonomy and Intelligent Distributed Systems (SAIDS) Lab, led by Lars Lindemann, focuses on systems and control theory, formal methods, and autonomous systems safety. The Learning and Interactive Robot Autonomy (Lira) Lab, led by Erdem Biyik (a former postdoc at UC Berkeley's Center for Human-Compatible AI), works on robot learning, human-robot alignment, and safe human-robot interaction, and has a trial collaboration with the Berkeley Existential Risk Initiative (BERI). The FORTIS Lab, led by Yue Zhao, focuses on AI auditing, safety, and reliable AI systems, developing tools and benchmarks for evaluating the trustworthiness of foundation models and agent systems. The USC Information Sciences Institute (ISI) is a major research institute within the Viterbi School of Engineering with combined research expenditures exceeding $100 million. ISI participates in the US AI Safety Institute Consortium (AISIC) organized by NIST, with Adam Russell (ISI's AI division director) having previously served as Chief Vision Officer at the US AI Safety Institute. USC also hosts the Center for AI in Society (CAIS), a joint venture between the School of Social Work and the Viterbi School of Engineering focused on AI for social good. USC separately received a $170,000 grant from the Future of Life Institute in 2022 for its Hollywood, Health & Society program at the Annenberg Norman Lear Center, supporting educational initiatives about nuclear weapons risks and irresponsible AI uses including autonomous weapons. The university also has active AI Safety and Effective Altruism student organizations that run AI safety curricula and career advising programs.