Collin Burns started his new job at the Commerce Department on Monday. By Thursday, the White House had forced him out. His offense was not incompetence, misconduct, or a security violation. It was his former employer: Anthropic.

The Washington Post first reported that Burns, a PhD computer scientist from UC Berkeley who had previously worked at both OpenAI and Anthropic, was hired to lead technical evaluation work at the Commerce Department's Center for AI Standards and Innovation. The center, formerly the Biden-era AI Safety Institute, is responsible for assessing national security risks posed by frontier AI models.

Senior White House officials had not been briefed on Burns's selection before his start date, according to four people with knowledge of the situation. When they learned of his Anthropic background, the response was immediate. Burns was asked to resign.

"A Dumb but Predictable Own Goal"

Dean Ball, a former Trump administration AI adviser, did not mince words. He called the removal "a dumb but predictable own goal" on social media, noting that Burns had "given up stock options and relocated" for the position and was "rewarded by his country with a punch in the face."

The context is charged. President Trump has publicly called Anthropic a "RADICAL LEFT, WOKE COMPANY." The administration has clashed with the AI firm over military access to its systems. The two sides are currently in active litigation over the administration's decision to label Anthropic a national security risk.

The Talent Problem Nobody Wants to Solve

The firing exposes a structural problem that goes beyond political grudges. The pool of people qualified to evaluate frontier AI systems for national security risks is extraordinarily small. Almost all of them work at, or have worked at, Anthropic, OpenAI, Google DeepMind, or Meta. If prior employment at any of those companies is disqualifying, the government is left choosing from a talent pool that does not have the expertise to do the job.

Chris Fall, a scientist with a career spanning federal agencies and academia, replaced Burns. The Commerce Department said Fall "brings the scientific leadership needed to ensure America leads the world in evaluating frontier AI models." Whether someone without direct frontier AI lab experience can effectively evaluate those models is a question the statement conspicuously avoids.

Signs of a Thaw?

Oddly, the firing happened the same week the White House hosted Anthropic CEO Dario Amodei for talks. The relationship between the administration and Anthropic appears to be simultaneously improving at the executive level and deteriorating at the operational level. Amodei gets meetings. His former employees get fired.

The center Burns was hired to strengthen is tasked with assessing national security risks from AI systems, including Anthropic's Mythos model. The administration now needs that work done by people who have never built the systems they are supposed to evaluate. That is not a policy. That is a handicap.

First reported by The Washington Post. Additional details from Slashdot and CompleteAITraining.