DeepMind’s Demis Hassabis proposes a new regulatory body modeled after FINRA to oversee the release of frontier AI models.

DeepMind CEO Demis Hassabis has called for the creation of an independent standards body to oversee the development and release of frontier artificial intelligence models, arguing that existing frameworks are insufficient for the rapid pace of innovation.

Why a New Regulatory Model Is Needed

Hassabis likens the proposed body to the Financial Industry Regulatory Authority (FINRA), which supervises brokerage firms and protects investors. He believes a similar model could provide clear, enforceable standards for AI developers while allowing for flexibility as the technology evolves.

According to the DeepMind chief, the current patchwork of national regulations and voluntary industry guidelines fails to address cross‑border challenges and the potential for high‑impact AI systems to be released without adequate safety checks.

Key Features of the Proposed Body

  • Independent governance structure with representation from academia, industry, and civil society
  • Mandated pre‑deployment risk assessments for models deemed "frontier"
  • Authority to issue binding safety certifications and enforce compliance
  • Mechanisms for rapid information sharing about emerging risks

The standards body would also maintain a public registry of frontier AI models, providing transparency for regulators, researchers, and the broader public.

Industry Reaction

Several AI firms have expressed tentative support for a coordinated approach, noting that a clear regulatory pathway could reduce uncertainty and foster responsible innovation. However, some smaller startups worry that additional compliance costs could hinder competition.

Hassabis emphasized that the goal is not to stifle progress but to ensure that powerful AI systems are deployed safely, with mechanisms for accountability and remediation if harms arise.

We need a trusted, independent entity that can set and enforce standards, much like FINRA does for finance.

The proposal comes amid growing global concern over AI safety, with governments in the United States, Europe, and Asia exploring their own regulatory frameworks.

Read the report