Photo: President Donald J. Trump delivers remarks in the East Room of the White House. (Official White House Photo by Daniel Torok)

Trump's AI Accord Can't Fix America's 73% AI Distrust

Here's a number that should worry everyone who was in that room on Tuesday: 73%.

That's how many Americans told Reuters/Ipsos they don't think AI companies have done enough to keep them safe. More than half — 55% — said we'd be better off slowing the whole thing down.

And then? Twenty of the richest tech CEOs on earth walked into the White House, sat down with Donald Trump, and signed a one-page piece of paper promising to... police themselves.

You can't make this stuff up.

Update, Oct. 1: the pledge's signatories are warning investors about the same risks

Two developments since publication sharpen the story. First, Anthropic's own IPO filing warns investors that poorly managed AI "could be a risk to humanity as a whole" and could outstrip industrialization in its economic imprint — even as the company seeks a $2 trillion valuation (Reuters, September 28). The same CEO who signed a "morally binding" pledge is telling investors his own technology might be an existential risk. Second, at the same-day America.gov launch, Trump claimed the new government AI chatbot "could not in theory be hacked into" by bad actors — without offering evidence (Straits Times, September 29).

A "Constitution" Nobody Has to Follow

Trump called the document "morally binding." Let that sink in for a second. Not legally binding. Morally binding.

"It's almost like a constitution, in a way," he said, standing in the East Room. "And the biggest people in the world signed that, and I signed it as president."

The paper itself — titled "Joint Commitment on Frontier Responsibilities" — asks six AI labs to do four things: set up internal controls, hire an internal team to check those controls, bring in an outside auditor, and have their boards read the reports. That's it. That's the whole thing.

Oh, and there's a line saying "over time, it may make sense to codify these steps into laws." May. May make sense. Someday. Maybe.

I've read a lot of tech policy documents. This one reads like it was designed to sound serious while committing to absolutely nothing enforceable. Because that's exactly what it is.

Who Was Actually There

Six CEOs signed: Sundar Pichai (Google), Dario Amodei (Anthropic), Mark Zuckerberg (Meta), Greg Brockman (OpenAI), Jensen Huang (Nvidia), and Elon Musk (xAI). Trump sat between Musk and Huang. Make of that seating chart what you will.

But here's what caught my eye — the room was bigger than the signatures. Jeff Bezos was there. Satya Nadella. AMD's Lisa Su. Palantir's Alex Karp. Roughly twenty executives total, plus House Speaker Mike Johnson, who called it "a historic meeting."

So the entire AI supply chain showed up. Chips, cloud, models, data. Everyone who makes money from this boom was at the table.

Know who wasn't? A single consumer advocate. A single labor rep. A single person from any of the communities getting a 500-megawatt data center dropped next to their kid's school. Not one.

Follow the Money

Quick detour, because context matters here.

AI executives and their investors were among the biggest individual donors to MAGA Inc. — Trump's super PAC — in 2025. OpenAI's Greg Brockman and his wife Anna alone gave $25 million, per FEC filings. Twenty-five. Million.

Does that prove anything corrupt happened in that room? No. But it tells you the vibe. These guys weren't hauled before Congress. They were hosted. Lunch at the White House. Cameras. Handshakes.

And what did the White House ask for in return? Basically: keep doing what you're doing, but write it down this time.

Even the Signatories Sound Nervous

This is my favorite part. Dario Amodei — the Anthropic CEO, a guy who signed the paper — couldn't even sound fully convinced.

"The technology has very real risks," he said. "And, you know, the mechanism, how we address those risks is still under discussion."

Still under discussion! The man just signed a safety commitment and in the same breath admitted nobody's figured out the mechanism yet. You've got to respect the honesty, even if it torpedoes the entire premise.

Outside the room, people were less diplomatic. Robin Jia, a computer science professor at USC, said the whole thing hinges on whether these companies' goals actually line up with the public interest — "particularly given recent hacking incidents." That's academic-speak for "have you seen what's been happening lately?"

And Alex Pascal, who runs Harvard's Berkman Klein Center, just went for it: "The only way to reduce the risks and harms of AI to Americans is through robust legal liability, regulation and fundamentally changing the race dynamics that are driving unsafe decision-making by frontier AI developers whose costs are borne by the American people."

Translation: self-policing is a joke, make them liable, change the incentives. Hard to argue.

The Data Center Bribe

Buried in the coverage was the other big topic: data centers. Towns across America are revolting — bipartisan, grassroots anger over these massive facilities eating farmland and spiking electricity bills.

Trump's answer? The companies will make communities "happy."

"You're going to see, data centres are going to be very popular," he said, surrounded by billionaires on the White House driveway. "These big, powerful, very rich, very smart companies are going to be making massive contributions to communities."

Very rich. Very smart. Very... reassuring? They'll fund some schools, maybe cut your power bill a little. Trust them.

Look, maybe they will. But "we'll make it up to you later" isn't a policy. It's a vibe. And vibes don't lower your electric bill.

What Actually Happens Next

Three things to watch — because the paper itself tells you almost nothing about enforcement:

One: Trump says he'll name a committee of about ten people to "watch over the whole enterprise." Who he picks tells you everything. Industry insiders? Theater. Actual critics with power? Interesting. My money's on theater, but let's see.

Two: a new White House AI overseer, "coming in days." Again — who, and with what authority? A coordinator with no teeth is just a press release with a salary.

Three: whether Congress ever turns any of this into law. Don't hold your breath. Speaker Johnson already said there's no need for "a moratorium" and warned that real regulation means "we're going to lose this race to China." That's not the sound of legislation coming.

The Bigger Picture

Same day, two other things happened that tell you where this is all going.

The White House launched America.gov — an AI chatbot for government services. And Trump signed an executive order telling every federal agency to stop saying "artificial intelligence" and start saying "Super Intelligence" instead.

Read those together with Tuesday's lunch and the message is unmistakable: AI is the future, the industry can handle itself, and Washington's job is to speed things up — not slow them down.

Meanwhile, 73% of the country is sitting there going: slow down, please.

That gap — between twenty nodding executives in the East Room and three-quarters of Americans who don't buy it — that's the real story. Not the paper. Not the handshakes. The gap.

We'll see which side wins.

Trump AI trust-poll FAQ

What did the Reuters/Ipsos poll actually find? Conducted September 17-20, 2026, it found 73% of Americans worry AI companies have not done enough to prevent AI from causing serious harm to society, and 55% said slowing AI development would be a good thing.

What does the accord require companies to do? Four voluntary steps: maintain robust internal controls, run an internal team to check those controls, hire an independent external auditor, and have a board committee review the reports. It is "morally binding," not legally binding — there are no fines or enforcement powers.

Who signed it? Trump plus six AI leaders: Sundar Pichai (Google), Dario Amodei (Anthropic), Mark Zuckerberg (Meta), Greg Brockman (OpenAI), Jensen Huang (Nvidia), and Elon Musk (xAI). Roughly twenty executives attended in total.

What happens next? Trump floated a roughly ten-person oversight board and promised a new White House AI policy official "in days" — but the administration has not identified potential appointees or defined the board's statutory authority. Speaker Johnson has signaled no appetite for binding legislation.

Sources

We pulled this together from CNN, Fast Company, Reuters, Le Monde, TBS News, TechX Media, Cointelegraph, and the White House's own Truth Social posts. FEC donation figures via Federal Election Commission records. Poll numbers from the Reuters/Ipsos survey conducted September 17–20, 2026.