OpenAI said Thursday it will comply with President Trump's executive order requiring frontier AI labs to submit their most advanced models for federal review before release, the first major lab to publicly accept terms Trump signed Tuesday and that the industry had spent months resisting.

The commitment lands in a week in which Washington tightened its grip on the AI sector on three fronts: model safety review, public-market disclosure and chip-export policy.

What OpenAI agreed to

George Osborne, the former U.K. finance minister who now serves as OpenAI's head of countries, told CNBC at SXSW in London that the company would sign up to the voluntary order, which asks AI developers to hand over frontier cyber models for a 30-day federal assessment before release. The order also calls for a benchmarking process to determine which systems should be designated "covered frontier models."

"It's quite right that democratic governments have a big role to play in how this technology is used and deployed," Osborne told CNBC's Arjun Kharpal. He said OpenAI takes its responsibilities "very seriously" and had "proactively suggested ways that governments can keep a track on safety and security issues, not just in the U.S., but more broadly." The order was Trump's second attempt at the framework; he walked away from a nearly identical signing ceremony less than two weeks earlier.

Anthropic's IPO clock

Anthropic's confidential S-1 filing, lodged with the Securities and Exchange Commission on Monday, has begun to recalibrate private valuations across the sector. Anthropic ended May at a $965 billion valuation on a $47 billion revenue run rate, figures that will now meet public-market scrutiny for the first time.

PitchBook analyst Harrison Rolfes said the filing "starts the clock on what will be the most scrutinized public offering in tech history." The deciding number, he said, is gross margin: "No one outside Anthropic has ever seen [gross margin], and it will either validate or collapse the entire narrative the private markets have been pricing for three years." D.A. Davidson's Gil Luria said Anthropic "appears to have the lead" in frontier models but warned that much of its usage "is for trials and experimentation and that may not sustain."

Warren calls Huang

Sen. Elizabeth Warren, D-Mass., on Thursday invited Nvidia Chief Executive Jensen Huang to testify before the Senate Banking Committee on June 11 on the company's China sales and export-control posture, asking him to confirm attendance by Monday. Warren told CNBC's "Squawk Box" on Wednesday that the chips at issue are "not just chips to help the AI industry in general" but, in China, "are actually used for military purposes."

Counterpoint

Today's reporting came entirely from CNBC, a center-leaning business outlet. Warren's pressure on Nvidia surfaces the left-of-center critique that U.S. firms profit from sales that erode national security; an industry-side reading, that voluntary compliance with the Trump order amounts to regulatory capture by incumbent labs, did not appear in today's sources. Civil-liberties groups and AI-safety researchers were not quoted. Nvidia did not respond to CNBC's request for comment, and the White House had not publicly addressed OpenAI's announcement by press time.

Huang's deadline to confirm the Senate appearance falls Monday. Anthropic's prospectus is expected to be made public in the coming weeks.