Li Fei-Fei is a computer scientist and AI governance figure. She is the Sequoia Professor of Computer Science at Stanford University and co-director of the Stanford Institute for Human-Centered AI (HAI). She led the creation of the ImageNet dataset, served as Chief Scientist of Google Cloud (2017–2018) and director of the Stanford AI Lab (2013–2018), co-founded the spatial-intelligence startup World Labs, and co-led the Joint California Policy Working Group on AI Frontier Models that produced The California Report on Frontier AI Policy for Governor Newsom.
Background and roles
Li is the Sequoia Professor of Computer Science at Stanford University and co-director of the Stanford Institute for Human-Centered AI (HAI). She previously served as Chief Scientist of Google Cloud (2017–2018) and as director of the Stanford AI Lab (2013–2018).
In 2009, Li led the creation of the ImageNet dataset, which contains roughly 14 million labeled images across about 20,000 categories, and the annual ImageNet Large Scale Visual Recognition Challenge (ILSVRC). AlexNet's 2012 ILSVRC victory, widely credited with launching the deep learning era, was built on ImageNet.
In 2019, Li co-founded HAI with John Etchemendy and continues to co-direct it. HAI is an academic center spanning AI research, policy, and society. The Stanford HAI AI Index Report is produced under HAI.
In 2025, Li co-led the Joint California Policy Working Group on AI Frontier Models, which produced The California Report on Frontier AI Policy for Governor Newsom (June 2025). The report provides the academic evidence base for California's "trust but verify" approach to frontier AI governance.
World Labs
Li co-founded World Labs in 2024 with AI researchers Justin Johnson, Christoph Lassner, and Ben Mildenhall to build "world models" that process spatial data and develop a physics-based understanding of three-dimensional environments (Source: fastcompany.com). The company emerged from stealth in September 2024 and has raised $1.23 billion across two rounds; on February 18, 2026, it raised $1 billion in a round whose investors included AMD, Autodesk (anchoring with $200 million), Emerson Collective, Fidelity Management & Research Company, NVIDIA, and Sea (Source: theaiinsider.tech; startuphub.ai).
World Labs released its first model, Marble, in November 2025; it generates editable, persistent 3D environments from text, images, video, or 3D layouts, exports meshes and video for game and visual-effects workflows, and is offered in free and paid tiers (Source: fastcompany.com). Li describes the company's goal as "spatial intelligence," which she calls the frontier beyond language that links imagination, perception, and action (Source: drfeifei.substack.com).
Positions and statements
Li has been described as a practitioner advocating evidence-based governance that balances innovation with risk reduction, rather than a researcher focused on existential risk. Her stated positions include that AI policy should be grounded in empirical research and multidisciplinary analysis rather than ideology; that early design choices in governance create path dependencies; that transparency and third-party verification are central governance tools; and that whistleblower protections and adverse-event reporting systems are necessary complements to industry self-disclosure.
She has testified before Congress multiple times on AI policy. The "trust but verify" framing from the California Report influenced the approach California took in SB 53.
Relationships
- related: The California Report on Frontier AI Policy — co-lead of the Working Group that produced this report
- related: Stanford HAI — co-director
- related: Stanford HAI AI Index Report 2026 — produced under institution she leads
- related: Gavin Newsom — California Report was commissioned by Newsom's office
- related: California SB 53 — Transparency in Frontier AI Act — academic framework that SB 53's approach reflects
- related: World Models — World Labs builds world models for spatial intelligence