How Zuckerberg Shut Down the White House AI Regulator With One Phone Call

The Call That Changed AI Policy

Before any public hearing, before any stakeholder comment period, before the regulatory draft ever saw daylight, Mark Zuckerberg picked up the phone and called Donald Trump to object to a proposed AI regulator. That detail — the direct line from a tech CEO to the Oval Office — is the story.

The White House had spent weeks developing a proposal to create a new independent body modeled on FINRA, the self-regulatory organization that oversees U.S. securities markets. The idea came from Google scientist Demis Hassabis, a Nobel Prize winner who published an essay in July laying out his case for a FINRA-style standards body to vet advanced AI models for cybersecurity risks before they were deployed broadly. White House officials briefed the concept on Trump in mid-August and separately walked major tech companies through it, including Meta, OpenAI, and Anthropic.

Trump called Zuckerberg first the week of August 17. What happened next, according to a senior White House official familiar with the conversation who requested anonymity, was a quiet but firm opposition. Zuckerberg did not ask Trump to scrap the idea outright. He made a more pointed argument: the people the White House might appoint to such a regulatory body should reflect Trump’s own light-touch approach to AI, not the kind of precautionary framework Hassabis was proposing.

The result was not a formal policy reversal, but it was a signal. The regulator remains under consideration, but its shape has already been negotiated in rooms most Americans have never heard of.

Who Benefits From a Closed Door

The FINRA model was appealing on paper. It would create an industry-led watchdog reporting to a government body, staffed partly by regulators drawn from the companies they oversee. Over 3,000 financial firms fund it. The structure looks like oversight while ensuring oversight comes from inside the industry.

But AI is not finance. A FINRA-style body could review models for safety risks, but as one White House official noted, even FINRA has little say in approving new financial products before they reach consumers. The analogy breaks down where it matters most. Still, the optics of a self-regulating AI body were enough to generate interest among some administration officials.

What the public did not see was how quickly that interest got neutralized. David Sacks, Trump’s former AI and crypto czar, had already branded the proposal a “DMV for AI,” imagining models “lineed up in a queue, waiting to get their test done.” In May, Sacks pulled a similar move: he called Trump the morning of a planned signing ceremony and convinced him to delay a sweeping AI executive order. One phone call. That’s how informal policymaking works when the informal channel leads directly to the president.

Sacks has since pivoted to advocating for something softer still: an AI regulator modeled on the Motion Picture Association, the voluntary ratings body that assigns films “R” and “PG-13” labels. In an August podcast episode, he said the system has Elon Musk’s support and argued that the MPA approach once “forestalled more intrusive, heavy-handed government action.” Musk did not respond to a request for comment.

The China Calculus

Zuckerberg’s private argument carried a specific framing that deserves attention. In an August essay, he wrote that any policy slowing a model’s release — even by a month — would “add significant risk to American leadership” over China. It is a clean, competitive logic: slower regulation means slower deployment, and slower deployment means ceding ground to Beijing.

That argument is neither new nor unique to Meta. But the fact that it was delivered in a private call to the president, rather than through a public testimony or industry association white paper, changes the nature of the influence. Public advocacy invites scrutiny. Private advocacy does not.

It is also worth noting which companies have stayed conspicuously silent. OpenAI and Google did not respond to requests for comment on the regulator debate. Anthropic declined to comment, though co-founder Jack Clark posted favorably about the idea on X in July. Meta’s spokesperson declined to provide a statement about Zuckerberg’s call. None of these companies have publicly taken a position that would survive a hearing room. Their positions were formed in calls like the one Zuckerberg made.

What Happens Next

Two paths remain open inside the White House. One is the FINRA-style regulator with some government involvement. The other is the MPA-style voluntary body with far less. Both are industry-influenced structures. The real question is not whether AI gets regulated, but what flavor of self-regulation wins.

The risks of inaction are real. AI agents are now capable of browsing the web, operating software, and writing code with limited human supervision. There have been a string of high-profile incidents where these systems escaped human control. Used by hostile actors, the same tools could identify infrastructure vulnerabilities or automate cyberattacks at speeds that outpace existing safeguards.

But the risk of overreach is also concrete. A regulator modeled on financial oversight would import a framework designed for markets with capital requirements, disclosure rules, and enforcement mechanisms that do not exist in AI. The comparison was always imperfect. The closer the alignment with FINRA, the more likely the body becomes a venue for industry to manage its own narrative rather than a genuine check on capability.

The longer-term concern is structural. When tech CEOs can resolve or redirect major policy questions through a single phone call to the president, the incentive shifts away from building broad coalitions and toward cultivating private access. That is not a new dynamic in Washington. It is new here because AI policy is being written at a pace that leaves almost no time for institutional checks to catch up.

The White House stated that the administration is “committed to balancing innovation and security in AI policymaking.” The call between Zuckerberg and Trump proves the commitment is real. It also proves the balance is already tilted.