AI·November 9, 2026·2 min read

U.S. AI safety center begins reporting informally to White House

Leadership gaps and halted public research raise independence questions

AI neural network concept illustration.
IllustrationAI neural network concept illustration.No Spin Media AI Illustration (generated with Gemini)

The small federal office responsible for independently evaluating the safety of advanced artificial intelligence systems has been asked to informally report to a White House office, according to two former government officials familiar with the discussions, raising questions about how independent its work will remain.

The Commerce Department's Center for Advancing Innovation and Standards for Super Intelligence, or CAISSI, has also been operating without a permanent director and has stopped publicly releasing its AI safety research. The center was previously known as the U.S. Center for AI Standards and Innovation and traces its work to the federal AI safety institute created during the Biden administration.

The reporting arrangement is not the same as formally transferring the center to the White House. The former officials described ongoing conversations about informal reporting, and the White House did not answer specific questions from The Associated Press about the arrangement. It said President Donald Trump had established a new Super Intelligence Force and commissioned a report on AI risks through his Council of Advisors on Science and Technology.

Critics of the arrangement say independent government testing is particularly important because increasingly capable AI systems can create risks involving cybersecurity, biological weapons, privacy and consumer protection. Caroline Siegel Singh of the Federation of American Scientists said the goal should be to have nonpartisan technical experts making policy based on what best serves the public. Other observers argue that closer White House involvement could increase the center's influence and potentially bring additional resources.

CAISSI currently receives about $10 million annually, according to AP, roughly one-eighth of the budget of its British counterpart. Its leadership has also been unstable: a director appointed in April was fired within days, a replacement resigned after several months, and the center has remained without a permanent leader. Congress began questioning the center in August after it stopped publicly releasing reports about AI systems' safety performance.

The broader dispute comes as the Trump administration emphasizes U.S. dominance in AI while encouraging technology companies to undertake substantial self-policing of their systems. That leaves unresolved questions about how much independent testing the federal government will conduct as AI becomes more capable.

Community discussion
"How independent should federal AI safety testing be from both the White House and the companies developing the technology?"