Fei-Fei Li Advocates Independent Oversight for AI Safety Evaluations

AI pioneer Fei-Fei Li called for independent institutions and the public sector to establish standardized benchmarks for AI safety rather than relying solely on tech companies to evaluate their own systems. Speaking in a Bloomberg TV interview, she emphasized that government, academia, and industry must jointly share responsibility for AI governance. As frontier AI systems gain capabilities and move into real-world applications, relying exclusively on self-regulation by tech developers presents potential conflicts of interest. Li's stance strengthens the global movement toward transparent, multi-stakeholder standards that ensure safety without relying solely on corporate discretion. Li framed AI risks as a governance challenge similar to historical innovations like electricity and steam engines, asserting that humans must retain ultimate control. She also noted that evaluation challenges extend beyond current large language models to spatial capabilities, such as the 3D 'world models' being developed at her company, World Labs.