AI Labs Propose Oversight Body; Offer to Staff It, Fund It, and Define What It Oversees

WASHINGTON—Demis Hassabis, chief executive of Google DeepMind, proposed Tuesday that the world's most capable AI systems be governed by a self-regulatory body modeled on FINRA, the Financial Industry Regulatory Authority — an organization created by member firms in 2007 to oversee member firms and active, without interruption, throughout the twelve months preceding the collapse of the global economy.

Under the plan, frontier laboratories would voluntarily submit their most capable models to the body thirty days before public release, at which point the board could recommend development slowdowns. The board's membership, enforcement authority, appeals process, and definition of which models qualify as "frontier-class" are each to be determined once the body is constituted. The founding membership, Hassabis confirmed, would be drawn from entities currently in possession of frontier-class models, which he described as the "natural constituency."

"This gives government genuine authority to act," said a spokesperson for a company that would hold board seats in the body and release models subject to the body's review. "Voluntary cooperation from the largest players sends the strongest possible signal."

The thirty-day submission window, analysts noted, is shorter than the average embargo on a frontier model press release.

Dr. Claire Weston, a senior AI governance researcher who has provided expert comment on fourteen previous industry self-regulatory proposals, said she would submit written feedback during the public comment period, which the framework has not yet established a process for creating.

She said she planned to adapt her existing comments from the March proposal. Most of the language, she noted, would remain accurate.