The Lede · Independent Reporting Friday, May 21 · Issue 421

Today's Picks

Recommended For You

Politics

Science

FEATURE

Inside the Quiet Lobbying War Over the Next AI Executive Order

By Vittoria Elliott · May 21, 2026

The first sign that something had shifted came from a memo no one was supposed to see. By the time it surfaced on a Sunday night in March, the document had already passed through at least four separate hands inside the West Wing — each of them adding edits in tracked changes that read, in the end, like a slow-motion negotiation between rival factions inside the same administration.

Over the next seventy-two hours, lobbyists for three of the largest AI labs in the country would arrive at the same conclusion independently: a draft executive order on frontier-model safety was real, it was farther along than anyone had publicly acknowledged, and it would land within weeks. What none of them knew yet was how much of the underlying language had been written not by the White House but by a small policy shop in Arlington with deep ties to a venture firm.

"The fight isn't over whether to regulate. It's over which company gets to write the regulation."

That trajectory — from a procedural memo to a coordinated, multi-million-dollar influence campaign — is the closest thing the modern AI industry has produced to a Washington power play. It has pulled in former senators, retired generals, and at least one sitting cabinet member's chief of staff. And it has, according to people familiar with the discussions, badly fractured the coalition of labs that two years ago stood shoulder-to-shoulder in a Senate hearing room promising to govern themselves.

What follows is a reconstruction of how that fracture happened, drawn from interviews with seventeen current and former government officials, lobbyists, and lab employees, most of whom spoke on the condition of anonymity to describe sensitive negotiations.