Fei-Fei Li calls for independent oversight of AI safety

Fei-Fei Li, the AI pioneer and co-founder of World Labs, called Tuesday for greater oversight of artificial intelligence by independent and public-sector entities, arguing that safety assessments of increasingly capable systems should not be left solely to the companies that build them.bloomberg

Speaking on Bloomberg TV, Li said that while researchers inside AI companies can measure the power and security attributes of individual models, that is no substitute for broader standards developed by governments, industry, and academic institutions. She argued that independent benchmarks, academia, government, and industry all have roles to play in evaluating increasingly powerful systems.bloomberg+1

A Consistent Message Amid a Shifting Landscape

Li's remarks align with a position she has staked out over the past year. At the Ai4 conference in Las Vegas in August, she appeared alongside Geoffrey Hinton and Andrew Ng in a debate over AI risks, where she advocated for sector-specific regulation and public investment rather than sweeping AI-specific restrictions. She has also urged policymakers to update existing regulatory frameworks in healthcare, transportation, finance, and education to account for AI, rather than attempting to "put the genie back in the bottle".forbes+1

In testimony before the Senate Committee on Homeland Security and Governmental Affairs earlier this month, Li pressed for demystifying AI, safeguarding its use with privacy and fairness measures, and investing in public AI research.hai.stanford

A Broader Debate Over Self-Regulation

Her call arrives at a moment of intensifying scrutiny over whether the AI industry can police itself. Just last week, Forbes reported that Anthropic CEO Dario Amodei's proposal for self-regulation — in which companies would select their own external evaluators — drew criticism for not going far enough. Critics compared the idea to allowing airlines to choose their own safety inspectors.forbes

Meanwhile, the EU AI Act's high-risk system requirements take full effect in August 2026, and multiple U.S. states have introduced legislation requiring independent audits of AI systems. The NIST AI Agent Standards Initiative, launched earlier this year, is also developing standards for autonomous AI agents that operate without continuous human oversight.gunder+2

Li, who serves as CEO of World Labs and co-directs Stanford's Institute for Human-Centered AI, framed the issue as one of ensuring humans remain in control of AI's trajectory. "AI's future is about humans," she said in the Bloomberg interview.bloomberg