US AI Safety Watchdog Left Leaderless as Staff Report Informally to White House

The federal office charged with testing whether America's most powerful artificial intelligence systems are safe has no permanent leader, has stopped publishing its findings, and now reports informally to the White House, the Associated Press reported on Sunday, 8 November 2026. The arrangement has raised fresh questions about the independence of the one United States body whose job is to assess the safety of new AI systems before they reach the public.

Staff at the Center for Advancing Innovation and Standards for Super Intelligence, known as CAISSI, have recently been asked to report informally to an office of the White House, two former government officials told the AP on condition of anonymity. The Commerce Department division tests how safely advanced AI models behave.

Key Points

• The Associated Press reported on 8 November 2026 that staff at CAISSI, the Commerce Department's AI safety centre, have been asked to report informally to a White House office.
• The centre has no permanent leader and has stopped publishing its research on how safely AI systems perform.
• CAISSI grew out of the Biden-era US AI Safety Institute and was renamed the Center for AI Standards and Innovation in June 2025 before taking its current, longer name.
• Its budget is about $10 million a year, roughly an eighth of its British counterpart's, and its leadership has turned over repeatedly in 2026.
• Critics warn the informal White House line could politicise safety findings or mute them in deference to the tech industry.

What Happened

According to the AP report, the instruction for CAISSI staff to report informally to the White House is recent, and it comes when the centre is already diminished. It has no permanent director, and it has stopped publishing the research that once told the public how safely leading AI systems perform. The two former officials who described the arrangement spoke anonymously, and the White House did not answer the AP's specific questions about CAISSI.

Instead, the White House said President Donald Trump had formally established his new Super Intelligence Force, led by National Intelligence Director Jay Clayton, and had commissioned a risk report from his Council of Advisors on Science and Technology. Neither the White House nor the Commerce Department has explained who at CAISSI reports to whom, or whether the arrangement is temporary.

Background/Context

CAISSI began life as the US AI Safety Institute, set up under the Biden administration as the government's hub for testing advanced AI models. Its task was to evaluate frontier systems for dangerous capabilities, develop safety standards and share findings with policymakers and the public. It was renamed the Center for AI Standards and Innovation in June 2025, and has since taken its current, longer title.

That mission has been interrupted by a year of upheaval. In August 2026, Congress asked why the centre had stopped releasing its safety reports, after administration officials told it in June to pause public reports on its reviews of AI models. No public safety assessments have followed. At about $10 million a year, CAISSI runs on roughly an eighth of the funding of its British counterpart.

US AI Safety Watchdog: A Centre Without a Leader or a Public Record

The leadership question is the most immediate. Collin Burns, a former researcher at the AI company Anthropic, was pushed out as director after just four days in April 2026. His replacement, Chris Fall, resigned after about three months, on 20 July. The National Institute of Standards and Technology director, Arvind Raman, is now acting director of the centre.

Three leaders in a few months, none of them permanent, has left the unit without a stable head. The publication freeze compounds the problem: when a safety body stops publishing, the public loses its only window into what the government knows about the systems it is testing. Companies still release their own safety assessments, but those are self-assessments; independent verification is what CAISSI was built to provide.

Reactions/Quotes

Critics of the new arrangement were blunt about the risk. The Federation of American Scientists and the Brennan Center for Justice warned that an informal reporting line to the White House could politicise the centre's findings or mute them in deference to the tech industry, whose companies build the systems under review.

The concern is structural: a safety assessment that must pass through the political office it is meant to inform can be delayed, softened or shelved. Others take a different view, arguing that a direct line to President Trump could raise the centre's profile inside the administration and improve its chances of more funding.

Why It Matters

The stakes are practical. AI models are now embedded in products used by hundreds of millions of people, and the gap between what a model can do and what the public knows about its risks is wide. Independent testing is the main tool governments have to close that gap before a system is widely deployed.

Without it, the public must rely on the companies that build the models to grade their own work, and on foreign testing bodies, such as the British institute, to flag problems. That weakens the United States' ability to set safety standards, negotiate international rules or respond quickly when a new capability emerges. It is also a test of how the Trump administration governs AI.

What Happens Next

Several questions now hang over the centre. The first is leadership: whether the administration names a permanent director, and whether that person is a safety specialist or a political appointee. The second is publication: whether Congress, which asked about the missing reports in August, presses the administration to resume public safety assessments. The third is the Super Intelligence Force, which may become the administration's main vehicle for AI risk work.

Frequently Asked Questions

What is CAISSI? The Center for Advancing Innovation and Standards for Super Intelligence, the Commerce Department division that assesses the safety of new AI systems. It grew out of the Biden-era US AI Safety Institute.

Who leads CAISSI now? It has no permanent leader. NIST director Arvind Raman is acting director after Collin Burns was pushed out after four days in April 2026 and Chris Fall resigned after about three months on 20 July.

Why did CAISSI stop publishing its safety reports? The administration told the centre in June 2026 to pause public reports on its reviews of AI models. Congress asked about the halt in August 2026, and no public assessments have followed.

What is the Super Intelligence Force? A new body formally established by President Donald Trump and led by National Intelligence Director Jay Clayton, with a risk report commissioned from his Council of Advisors on Science and Technology.

Why does the White House reporting line matter? Critics, including the Federation of American Scientists and the Brennan Center for Justice, warn it could politicise safety findings or mute them in deference to the tech industry.

Read more on US News Pulse: China AI Law Speeds Up to Cover DeepSeek, Alibaba and Moonshot AI Models and Trump Backs Nvidia CEO Jensen Huang Amid Reported China Spy Plot Inquiry and Hidden Deepfake Apps Exposed: 400 Ordinary AI Apps Concealed Abusive Tools.