DeepMind CEO Demis Hassabis Calls for Global AI Watchdog — Led by the US
Google DeepMind CEO Demis Hassabis is calling for a US-led global AI watchdog with the power to evaluate frontier models before release and coordinate industry-wide slowdowns.
The Vision for a Global AI Regulator
Google DeepMind CEO Demis Hassabis has proposed creating a global AI watchdog with the authority to evaluate frontier models before their release. In a blog post, he argued the organization should be US-led, citing America's economic and technical standing as the world's AI leader.
How It Would Work
The proposed organization would resemble FINRA (Financial Industry Regulatory Authority). It would be composed of leading independent experts, industry representatives, and open-source community members. Core functions include pre-release evaluation of frontier models, coordinating industry-wide slowdowns if models are too dangerous, and establishing global standards.
Why Now
AI technologies are developing so rapidly that traditional regulatory approaches can't keep pace. GPT-class models, multimodal AI, and agentic systems are advancing faster than any government can legislate. The window for proactive governance is narrowing.
The US Leadership Question
Hassabis argues the US is best positioned to lead this initiative. He hopes the organization could be operational before the end of this year — an ambitious timeline reflecting the urgency of the risks.
Challenges and Criticism
Critics question whether one organization can effectively govern global AI development. Key challenges include: geopolitical tensions between US and China, industry resistance to regulation, rapid technological change outpacing any rules, and enforcement complexity across jurisdictions.
Georgia Perspective
For Georgia, global AI regulations matter as the country aims to become a tech hub. Aligning with international AI safety standards will help Georgian companies integrate into global markets and attract foreign investment.