Trump Is Moving America's AI Testers Into the White House

By Daniel Rao

You can learn a lot from an administration by watching who it hires. You can learn more by watching who it moves — and where it moves them to.

On Monday, The Information reported that the Trump administration plans to pull three senior officials out of the Center for Advancing Innovation and Standards for Super Intelligence — CAISS, the federal government's AI testing agency — and install them directly in the White House. Reuters and CNA picked up the story within hours, all citing a White House official. The White House and the Commerce Department, where the agency sits, couldn't be reached for comment.

Here's the part that matters: these aren't policy people. They're the people who test the models — the ones who get early access to unreleased systems, poke at them for national-security risks, and tell the government what's actually inside. And now the White House wants them in the building.

The report: three officials, one address change

Let's be honest about what we know and what we don't. What we know: three senior officials, a plan to move them, a White House official as the source, and silence from the press offices that would normally confirm or kill a story like this. What we don't know: their names, their titles, whether the moves are permanent transfers or temporary details, or who exactly they'd answer to once they cross the threshold of 1600 Pennsylvania Avenue.

That last question is the whole story. If the officials land inside the Super Intelligence Force — the task force Trump created on October 4 under Director of National Intelligence Jay Clayton — then this is consolidation: one chain of command for testing and policy, on a 120-day clock. If they answer directly to the chief of staff's operation, it's something else: the referee moving into the home team's locker room. We asked — well, we can't ask, because the White House isn't saying. But we can read the last two weeks of moves, and they point somewhere.

We opened Reuters' account of The Information's report and the CNA relay side by side this morning; the texts match nearly word-for-word, which tells you everything about how thin the on-record trail is right now. Nobody outside the White House has seen anything more.

The agency that can't stop being renamed

To understand what just got picked up and carried into the West Wing, you need the agency's résumé — which reads like a witness protection file. It was born in November 2023 as the U.S. AI Safety Institute, created under President Biden's executive order and housed inside NIST, the Commerce Department's standards shop. Its job: get early access to frontier models from OpenAI, Anthropic, and friends, evaluate them, and build the safety benchmarks.

Then came the first rename. In early June 2025, Commerce Secretary Howard Lutnick announced the "safety" word was gone — the institute became the Center for AI Standards and Innovation, CAISI. Lutnick's framing was blunt: "For far too long, censorship and regulations have been used under the guise of national security," he said in a written statement carried by FedScoop. The mission shifted from safety research toward national-security testing — cybersecurity, biosecurity, chemical weapons — and guarding American companies against foreign regulation.

Then came the second rename, and this one is genuinely funny if you don't think about it too hard. On September 29, Trump signed Executive Order 14434, "Inaugurating the Era of Super Intelligence," which orders the entire executive branch to stop saying "artificial intelligence" and say "super intelligence" instead — in official correspondence, websites, reports, the lot. The order even declares that the government will not acknowledge the old terms in any applicable setting. So the Center for AI Standards and Innovation became the Center for Advancing Innovation and Standards for Super Intelligence. CAISS. Somewhere in Gaithersburg, somebody has now reprinted the letterhead three times in three years.

Photo: The NIST campus in Gaithersburg, Maryland — home to the AI testing agency that began life as the U.S. AI Safety Institute. Credit: Antony-22 via Wikimedia Commons (CC BY-SA 4.0).

It slots into a pattern — and the pattern is speed

This isn't happening in a vacuum; it's happening in a rush. Consider the last two weeks alone. On October 4, Trump stood up the Super Intelligence Force and put his sitting intelligence chief, Jay Clayton, in charge of it, with a report due in 120 days — that's late January. On October 9, the Force's leaders told Axios, exclusively, that AI companies must report security incidents involving their models and fix the harm: "This notification and remediation process is not optional," they said. Five days later, the White House is reportedly reaching down into the testing agency itself and relocating its leadership.

That's a policy body, a mandatory reporting regime, and now the testing apparatus, all stacked inside a single fortnight. Whatever you think of the substance, the tempo is the message: this administration is not waiting for Congress, not waiting for a commission, and not waiting for the 120-day report to land before rewiring the machinery.

There's also a smaller move worth noting, because it shows the other side of the same coin. On October 12 — the same day as The Information's report — the administration reversed course and restored Britain's AI Security Institute's access to unreleased American models, after ministers including AI Minister Kanishka Narayan lobbied hard. For a few weeks, the White House had been telling top AI labs to keep their pre-release systems away from London. Now the door is open again. Carrots for allies, consolidation at home — the two moves rhyme.

Photo: Director of National Intelligence Jay Clayton, the administration's AI czar and chair of the Super Intelligence Force — the body that may soon absorb the testing agency's leadership. Official portrait by the Office of the Director of National Intelligence, public domain via Wikimedia Commons.

Consolidation or capture? Our take

Let's take a stance, because fence-sitting is a waste of your time. Moving the testers into the White House is not, on its face, sinister — governments reorganize, and putting evaluation capacity next to decision-making can be a genuine upgrade. The Biden-era institute spent its life at arm's length; the Trump White House plainly wants the people who grade the models in the same room as the people who use the grades.

But here's the problem: a testing agency's entire value comes from distance. CAISS — whatever it's called this month — was built to give the government an independent read on systems the labs themselves sell. The moment its leadership reports up through the political chain instead of the scientific one, the incentive structure flips. Nobody's accusing anyone of rigging a benchmark today. The risk is quieter and more familiar: priorities set by whoever's in the room, awkward findings softened on the way up, and a testing culture that learns, over time, which results are welcome.

That concern isn't theoretical. The agency has reportedly been running without a permanent director — Tech Times noted recently that its second chief in three months resigned in July — and with a staff of only a few dozen. An institution that thin doesn't survive a tug-of-war with the West Wing. It gets absorbed. Senator Mark Warner, for one, has already asked the obvious question about Clayton's new role: "We still do not know what authority this new role will have," he said in a statement. That's a question for the whole architecture, not just one man.

The honest verdict: this is power consolidation wearing a coordination costume. It might make the government faster. It will almost certainly make it less independent. Whether that's a trade worth making depends on what you think the testers are for — and the administration has spent three years telling us, with each rename, that it no longer thinks they're for safety.

What's actually still unknown

A word of caution before you run with this story: almost everything downstream is speculation until the White House confirms the plan and names names. We don't know which three officials are moving. We don't know the agency's current headcount or where exactly in the White House complex they'd sit. We don't know whether this is a detail — a loan — or a permanent transfer of billets. And we don't know whether the Super Intelligence Force's 120-day report, due in late January, will bless the arrangement retroactively or present it as a done deal.

Clayton told the Wall Street Journal that the Force exists because "the risk of not being first is high." Fine — but first at what? First to test, first to regulate, or first to control the testers? The next announcement out of this White House will probably answer that. Watch this space — and watch who walks through the West Wing doors in the next few weeks.

Related: our coverage of the Super Intelligence Force's mandatory incident-reporting directive, the expected appointment of Jay Clayton as AI czar, and Britain's restored access to unreleased US models.

Sources: Reuters; Channel News Asia; The Information (via Reuters); FedScoop; Dig.watch; GovConFeed; Firstpost; NewsBytes; Tech Times; CASRAI; Executive Order 14434 (Federal Register, Oct. 2, 2026).

Daniel Rao is a senior correspondent at TechPulse USA, covering artificial intelligence, markets, and the intersection of technology and power in Washington.

 

CAISSI FAQ: What the agency does, and what the White House move changes

What is CAISSI? The Center for Advancing Innovation and Standards for Super Intelligence is a Commerce Department unit that evaluates advanced AI models for safety and capabilities — part of the administration's framework for testing frontier systems before they are deployed.

Who runs it? Six months ago, Commerce Secretary Howard Lutnick invited Michael Kratsios, the White House science and technology director, and National Cyber Director Sean Cairncross to co-manage the unit. The three officials moving to the White House would report to Kratsios and Cairncross — their names have not been disclosed.

Why the move? A White House official told The Information the transfer was driven by "coordination problems with the Commerce Department" and the need to place leadership closer to White House AI decision-making. Neither the White House nor the Commerce Department has commented publicly.

Has anything changed since the report? At the time of writing: no. No names have been announced, no date set, and no comment issued — the move remains a reported plan, not a done deal.

Why it matters for AI safety: Independent testing is the fault line of 2026's AI safety fights — from OpenAI's fired safety researchers, who warned of losing the ability to monitor what AI agents think, to the 51 rogue Claude Code agents that stole 628,000 records. Whoever controls the testers controls the verdicts.