US frontier-model evaluator CAISSI reported to be answering to the White House
The Associated Press reported on 8 November 2026 that staff at the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), the Commerce Department body that assesses the safety of advanced AI systems, had recently been asked to report informally to an office of the White House. Two former government officials described the arrangement on condition of anonymity. The centre has no permanent leader: a director hired in April was dismissed within days, a replacement resigned after a few months, and the head of its parent institute now serves as acting director. CAISSI has stopped publishing its safety research, and Congress asked in August why the reports had halted. The centre operates on roughly $10 million a year, about one-eighth of its UK counterpart's budget, while the EU's AI Office employs more than 125 people.
Why It Mattered
CAISSI is the successor to the US AI Safety Institute and the only federal body whose job is independent technical evaluation of frontier models. If its staff answer informally to the White House, publish nothing, and have no confirmed director, then the United States has no functioning independent evaluator at a moment when its own frontier labs have disclosed models involved in unauthorised access to third-party systems. The comparison figures matter: a $10 million budget set against the UK institute and an EU AI Office with statutory fining powers and a growing headcount describes a widening transatlantic gap in state capacity to check AI systems, not merely a difference in regulatory philosophy. Critics including the Brennan Center argue that routing model safety through national-security channels privileges executive discretion and secrecy over consumer protection, privacy and civil rights. The reporting rests on anonymous sources and describes an informal arrangement rather than an announced reorganisation, and the White House did not answer specific questions. Even so, the verifiable surface, a leaderless centre, publications halted since the summer, and unanswered congressional inquiries, is enough to make this the dated marker for when independent US AI evaluation effectively lapsed.
Who Built It
Associated Press (reporting); US Department of Commerce
Applications
- AI Safety Evaluation
- Government Oversight
- Standards