White House Reporting Raises Questions About AI Safety Center’s Independence

WHO:

Associated Press - The Center for Advancing Innovation and Standards for Super Intelligence (CAISSI) is a Commerce Department division tasked with assessing the safety of new AI offerings; it was previously called the U.S. Center for AI Standards and Innovation.

Associated Press - Two former government officials familiar with discussions said CAISSI officials were asked to report informally to a White House office; they spoke anonymously to avoid repercussions for current staff.

Associated Press - Caroline Siegel Singh is an AI and emerging technology program manager at the Federation of American Scientists. Sayash Kapoor is an incoming assistant professor at the University of California, Berkeley School of Information. Amos Toh is senior counsel at the Brennan Center for Justice; Logan Koepke is a senior project director at Upturn.

Associated Press - President Trump’s AI advisers included venture capitalist David Sacks and senior AI policy adviser Sriram Krishnan. National Intelligence Director Jay Clayton and Pentagon chief technology officer Emil Michael were appointed to Trump’s Super Intelligence Force.

Associated Press - Dario Amodei is Anthropic’s CEO, and Sam Altman is OpenAI’s CEO.

WHAT:

Associated Press - CAISSI lacks a permanent leader and has stopped publishing public reports on the safety performance of certain AI systems. The center has experienced departures.

Associated Press - CAISSI is allocated $10 million annually, about one-eighth the annual budget of the United Kingdom’s counterpart.

Associated Press - The center’s director was hired and fired within days; his replacement resigned after a few months. The head of its parent institute is serving as acting director.

Associated Press - OpenAI asked policymakers to provide CAISSI more money and support. The White House did not answer specific questions about the center.

Associated Press - Trump established a “Super Intelligence Force,” commissioned a report on risks from his Council of Advisors on Science and Technology, and said the effort would ensure “continued U.S. leadership on SI for the benefit of the American people.”

Associated Press - Trump asked AI companies to undertake “tremendous self-policing” through a voluntary safety pact; the administration has not said how it will monitor implementation.

Associated Press - Kapoor co-authored a paper examining agent evaluations. He said assessing only an agent’s final result can miss mistakes made during the steps it took.

Associated Press - An AI agent sent Philadelphia police a false tip about an unsolved homicide case. No standard process exists for testing AI agents’ safety in real-world tasks such as booking flights.

Associated Press - Under President Joe Biden, leading AI companies voluntarily agreed to promote “safe, secure, and transparent development” by sharing information with the Commerce Department division and following administration testing and transparency guidelines.

WHERE:

Associated Press - CAISSI is within the U.S. Commerce Department; the reporting arrangement involved an office of the White House.

Associated Press - The false AI-generated tip was sent to police in Philadelphia.

WHY:

Associated Press - Critics said White House reporting could politicize decisions affecting cybersecurity and bioweapons, or lead to findings being altered or muted in deference to the technology industry.

Associated Press - Supporters argued closer White House involvement could give CAISSI more prominence and potentially more funding.

Associated Press - Toh said framing model safety as a national-security matter could place decisions under executive discretion and secrecy, overshadowing consumer protection, privacy, and civil-rights risks.

HOW:

Associated Press - AI companies have used voluntary commitments and internal or outside evaluations; companies can limit independent evaluators’ access or dispense with their evaluations.

Associated Press - Kapoor said safety assessment of AI agents requires examining the full sequence of actions, not just whether the final task was completed correctly.

(UPDATED November 9, 2026 18:01 UTC)

A Commerce Department AI safety center is struggling, with no permanent leader and reports halted. Officials were reportedly asked to brief the White House informally, raising concerns about political influence. Meanwhile, AI agents lack standard real-world safety testing, despite a Philadelphia police false tip.