EXPLORINEER
en

Archive — US — techcrunch

DeepMind CEO calls for an independent standards body to regulate frontier AI

DeepMind CEO Demis Hassabis proposes an independent standards body for regulating frontier AI, modeled after FINRA. This body would voluntarily review AI models before release, ensuring best practices and addressing vulnerabilities, amid criticism of current government assessments. The initiative seeks to balance technical rigor with innovation support, aiming to adapt to evolving AI risks.

Key points

  • Hassabis calls for a regulatory body for frontier AI akin to FINRA.
  • The body would review models up to 30 days before their release.
  • It aims to establish best practices and address post-release vulnerabilities.
  • Current U.S. government reviews of AI models face criticism for lack of expertise.
  • The proposed body would be funded by AI labs and operated independently.
  • Hassabis argues it would support innovation while managing risks.

Why it matters

  • The tech industry is currently debating the need for AI regulation, with varying viewpoints.
  • In 2025, AI safety and regulation trends highlight a shift towards self-regulation as seen in finance.
  • The initiative responds to growing concerns about AI accountability and effectiveness.
  • Ongoing discussions about tools and frameworks for AI governance reflect societal demand for safety.

Hassabis's proposal for a standards body represents a significant step toward establishing a regulatory framework in AI, aiming to enhance accountability while fostering innovation in a rapidly evolving field.