Demis Hassabis Proposes US-Led AI Standards Body
Demis Hassabis, the Nobel laureate behind Google’s AI, has warned that artificial general intelligence (AGI) is "a few short years away" and the world is not ready. In response, he has proposed a US-led Standards Body modeled after Wall Street’s FINRA to regulate the release of frontier AI models.
Key Points:
-
Regulation Before Release: The body would test powerful AI models before they are made public, focusing on potential risks like cyberattacks, biological and nuclear hazards, and deception.
-
Public-Private Partnership: Funded primarily by AI labs themselves and staffed with independent experts, including Turing Award winners, the body would operate under government oversight.
-
"Frontier-Class" Models: Once established, models would need approval before entering the US market, with benchmarks updated quarterly. Startups and academics would be exempt.
-
Brake Mechanism: The Standards Body could coordinate a slowdown in AI development if necessary to mitigate growing risks.
Hassabis argues that today’s cyberrisks are only "warning shots," expecting far more dangerous bio- and nuclear-capable tools within 18 months. He has already briefed the Trump administration, rival labs, and European officials on his proposal, receiving positive responses.