The richest people in tech just vetoed the President's AI policy—and the fracture lines inside MAGA matter more than the decision itself.
The Summary
- Trump shelved an AI oversight executive order hours before signing, after former AI czar David Sacks, Elon Musk, and Mark Zuckerberg pushed back. Trump cited fears it would "slow US efforts to beat China."
- The draft order emphasized voluntary reviews—developers could submit advanced models to federal agencies 90 days before release, but weren't required to. Still too much for the billionaires.
- The real story: much of the MAGA base supports AI regulation, creating a rare split between populist voters and their tech-baron allies.
The Signal
The seven-page draft went out of its way to be voluntary. It explicitly stated nothing in the order "shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement." Developers of frontier models—the ones that could "unleash devastating cyberattacks" if they fell into the wrong hands—could choose to let federal agencies review them. Up to 90 days before public release. No requirement. No penalty for skipping it.
That still wasn't enough. Sacks, Musk, and Zuckerberg convinced Trump to pull the plug hours before the scheduled signing. Trump's public explanation leaned on the China card: anything that slows down American AI companies helps Beijing. The subtext is simpler. The people building the models don't want anyone looking over their shoulder, even voluntarily, even with 90 days of runway.
"Nothing in this section shall be construed to authorize mandatory licensing—yet the order still died on contact with Silicon Valley."
This is where it gets interesting for Web4. The draft order wasn't about stopping innovation. It was about creating a review pathway for the models most likely to power autonomous agents at scale—the ones that will run financial systems, coordinate supply chains, make hiring decisions. The voluntary framework would have let companies signal safety to enterprise buyers and government contractors without giving up speed. A trust badge, basically, for the agent economy.
But the bigger companies don't want a trust badge system. They want no system. Because a voluntary review pathway becomes the industry standard, and then the standard becomes the expectation, and then you're explaining why you didn't submit your model when everyone else did. Voluntary review kills the "move fast" advantage for whoever opts out.
Here's the split no one's talking about yet:
- Tech billionaires: AI regulation = slower profits, Chinese advantage
- MAGA base voters: AI regulation = protect American workers from job displacement
- Enterprise buyers: want some kind of safety validation before deploying autonomous agents
Fortune reports that much of the MAGA coalition actually supports AI oversight, which puts the populist wing at odds with the donor class. That's the real fracture. The executive order died because three billionaires had more sway than the base. That math doesn't hold forever.
The Implication
Watch what happens when the first major autonomous agent causes real damage—a trading bot that tanks a market, a hiring agent that generates a discrimination lawsuit, a coordination system that fails during a crisis. The voluntary review framework will come back, probably with less voluntary language. The companies that killed this order just bought themselves 12 to 18 months to ship without oversight. They better use it to build safety in, because the next version of this won't be a 90-day optional review.
For anyone building in the agent economy, the message is clear: there's no federal safety standard coming soon, which means enterprise customers will create their own. If you're selling B2B agent infrastructure, start documenting your safety testing now. You'll need receipts when the corporate buyers start asking.