Skip to main content

Hassabis's AI Safety Vision: From DeepMind CEO to Industry Watchdog?

In the weeks before he handed over the CEO reins at Google DeepMind, Demis Hassabis wasn't just packing up his desk. According to the Wall Street Journal, the AI pioneer was quietly laying groundwork for something bigger: an independent body to oversee AI safety, modeled after the International Atomic Energy Agency (IAEA).

Hassabis, who co-founded DeepMind and has long been a voice for responsible AI development, reportedly discussed the proposal with senior Trump administration officials, including Treasury Secretary Scott Bessent and tech advisor Michael Kratsios. The idea? A non-governmental organization that would promote safety standards and best practices for general artificial intelligence—much like the IAEA does for nuclear energy.

A Google spokesperson confirmed that the company sees ensuring the safety of advanced AI models as "crucial for society," and noted that Hassabis is uniquely positioned to drive this mission forward. In his new role as Chairman of DeepMind and Chief Scientist at Alphabet, he'll focus on scientific breakthroughs and shaping the direction of AGI.

A 'Frontier-Level' Label for Safe AI

Back in mid-July, Hassabis took to X (formerly Twitter) to float the idea of an industry-funded "standardizing body" for AI safety. He suggested it could operate like the Financial Industry Regulatory Authority (FINRA)—a self-regulatory organization created during the Great Depression to oversee brokerage firms, funded by the financial industry itself.

Under his proposal, federal agencies and national labs under the U.S. Department of Energy would team up with AI companies to test models in areas tied to national security. Those that pass muster would earn a "Frontier-Level" certification—a sort of seal of approval for safe, reliable AI.

From Lab Leader to Systems Architect

Last week, Google announced a major shake-up at DeepMind: Hassabis stepped down as CEO, handing the reins to his deputy, Corinna Kretzschmar. But this isn't just a routine succession. It's a clear signal that Hassabis is pivoting from leading the world's most advanced AI lab to trying to build the guardrails for the entire industry.

Think about it: Hassabis is trading the role of technical trailblazer for that of a systems designer. If his proposed institution comes to life, it could reshape how AI is regulated globally. That's a big deal—and it's exactly the kind of move that makes you wonder what else he's been cooking up behind the scenes.

Key Points

  • Hassabis's Vision: An independent AI safety organization, inspired by the IAEA and FINRA, to set standards and certify safe AI models.
  • Government Talks: Discussions with U.S. officials, including Treasury Secretary Scott Bessent, took place before his CEO departure.
  • New Role: Hassabis is now Chairman of DeepMind and Chief Scientist at Alphabet, focusing on AGI direction and safety.
  • Potential Impact: If realized, this could become a cornerstone of global AI governance, affecting how AI is developed and deployed worldwide.