人工智慧先驱暨 World Labs 执行长李飞飞呼吁，应加强独立机构与公共部门对 AI 技术的监管，并强调 AI 系统的安全评估绝不能仅由开发该技术的私人企业主导。她指出，虽然企业内部研究人员可评估单一模型的性能与安全性，但政府、产业界与学术界仍须共同承担责任，建立更广泛、独立的评测基准。

李飞飞进一步指出，AI 的潜在生存风险本质上取决于人类如何开发、治理与使用该技术，就如同电力、蒸汽机与现代交通工具的发展历史一样，社会必须在技术利益与人类安全之间取得平衡。她强调「人类掌握科技的力量且必须保持控制」，AI 的核心不在于技术本身，而在于守护人类的集体与个人尊严、自主性以及社会责任。

面对矽谷高层对 AI 发展过速的担忧以及政界对生存威胁的不同看法，李飞飞创办的 World Labs 正在开发能重建三维环境的「世界模型」，应用范围涵盖机器人训练、设计与医疗保健等领域。她提醒，随著 AI 系统走向真实世界应用，整个产业正面临一个核心课题：如何精准评估能力日益强大的 AI，并确保其行为完全符合人类的预期目标。（关键数字：3）

Artificial intelligence pioneer and World Labs CEO Fei-Fei Li urged for stronger oversight by independent and public-sector entities, asserting that safety assessments of AI systems should not rely solely on the companies building them. She stated that while corporate researchers can measure the capabilities and security of specific models, broad standards and benchmarks collaboratively established by governments, industry, and academia remain indispensable.

Li emphasized that any existential risks posed by AI ultimately depend on human choices regarding the technology's development, governance, and application, drawing parallels to how society historical balanced benefits and safety with electricity, steam engines, and modern transportation. Asserting that humans own the power of technology and must stay in control, she noted that AI is fundamentally about preserving human dignity, agency, and responsibility rather than the technology itself.

Amid escalating debates among tech leaders and policymakers regarding extreme AI risks, Li's venture, World Labs, is pioneering world models that reconstruct 3D environments to power robotics, healthcare, and design applications. She highlighted that as these advanced systems transition into real-world deployments, the entire industry faces the fundamental challenge of rigorously evaluating increasingly powerful AI to verify that it behaves as intended.