A Push for Global AI Oversight Takes Shape
In a significant move for AI governance, Demis Hassabis, CEO of Google's DeepMind, has called for the creation of an international, independent body tasked with uniformly testing and reviewing advanced artificial intelligence models prior to their public release.
Framework and Proposed Authority
The envisioned body would operate autonomously, with its core mission being the in-depth scrutiny of cutting-edge AI models developed by leading companies. Review panels would consist of independent experts, while funding is expected to come from major AI firms. Under the proposal, models could undergo testing for up to 30 days.
A key aspect of the framework is the potential authority to coordinate the industry and, if deemed necessary, recommend a slowdown in the development pace of the most advanced systems. Hassabis highlighted the sophisticated cyber capabilities demonstrated by a recent AI model as evidence of the pressing need for stronger safety measures.
Industry Heavyweights Lend Support
The proposal has garnered swift public endorsement from two prominent figures in the tech world. Sam Altman, CEO of OpenAI, and Elon Musk have both expressed support for establishing an international safety review mechanism of this kind.
Hassabis is scheduled to travel to Washington, D.C. next week to discuss the plan in more detail with U.S. government officials, signaling a shift of the AI governance conversation from industry circles to the policy-making arena.