Zuckerberg Privately Called Trump to Oppose Establishing an AI Regulatory Body
nashnova research
Zuckerberg called Trump last week to push back against a proposed national AI regulator; the call shows Big Tech is bypassing normal policy channels to lobby the president directly, and the future of U.S. AI oversight is being shaped behind closed doors.
What did the call actually cover?
During the week of August 17, Trump initiated the call. Zuckerberg used it to explicitly oppose creating a national AI regulatory body.
Rather than go public, Zuckerberg told Trump privately: if a regulator must exist, appoint people aligned with Trump's own AI philosophy — widely read as pro-deregulation.
This means → Zuckerberg's play was not outright rejection but a fallback: control the appointments, dilute the oversight.
Who proposed this regulator and how would it work?
The driving force is Demis Hassabis, Google DeepMind scientist and Nobel laureate. In a July essay he proposed modeling a new AI "standards body" on FINRA — a self-regulatory organization funded by the financial industry to oversee broker-dealers.
Under the White House blueprint, the body would review and stress-test AI models before they go live. In plain terms = every major model would need a green light before release.
The White House briefed Trump and executives at Meta, OpenAI, and Anthropic on the plan in mid-August. The proposal has not been shelved despite Zuckerberg's objection.
What is the fight inside the White House?
Officials are weighing two paths: a FINRA-style regulator with review authority, or a voluntary industry body modeled on the Motion Picture Association (MPA) — the latter championed by David Sacks, Trump's former AI and crypto adviser.
Sacks called a government regulator "a DMV for AI" where models would "wait in line for approval." He favors voluntary labels for AI models, much like PG-13 or R ratings for films.
This reflects a fundamental split: should the government hold approval power, or should the industry label itself?
Where do the tech companies stand?
So far, not a single major AI company has publicly backed or opposed either regulatory proposal.
Anthropic co-founder Jack Clark expressed approval of the concept on X in July, but Anthropic declined to comment on its official position. OpenAI and Google did not respond.
This means → every company is watching the White House power struggle play out, unwilling to show its hand before the direction is set.
What is Zuckerberg really worried about?
In a public essay in August he wrote: any policy that could delay a model's release — "even by just one month" — would "pose a significant risk to America's lead over China."
In plain terms = Zuckerberg's logic runs: approval process = time cost = China catches up, so any form of pre-release review is unacceptable.
Critics counter that even "voluntary standards," once they become industry norms, are likely to be written into law and made mandatory — exactly the risk Zuckerberg wants to kill at the source.
Why does this matter?
In May, Sacks talked Trump out of signing the administration's first AI executive order with a single phone call on signing day — this means → one-on-one calls between tech executives and the president are already the most powerful informal variable in White House AI policy.
Carefully advanced internal proposals can be slowed or redirected at any moment by a phone call, leaving some senior officials feeling sidelined.
This reflects a larger reality: the direction of U.S. AI regulation depends not on policy research, but on who last had the president's ear.
市场有风险，内容仅供研究参考，不构成投资建议。