The White House’s request raises concerns about the independence of the US AI Security Center – Unite.AI

The Associated Press reported on November 8, 2026, that officials at the Center for Advancing Innovation and Standards for Super Intelligence, or CAISSI, had recently been asked to report informally to a White House office, raising concerns about the independence of federal AI safety assessments.

Two former government officials familiar with the ongoing conversations described the request on condition of anonymity to avoid repercussions on current staffers. The office was not identified and the reported discussions involve an informal agreement rather than an announced formal reorganization.

Critics have warned that the deal could politicize decisions involving cybersecurity and bioweapons, with outcomes potentially altered or diluted in deference to a tech industry largely relied on by police themselves. Caroline Siegel Singh, program manager for artificial intelligence and emerging technology at the Federation of American Scientists, has advocated for nonpartisan or bipartisan technocratic experts to make policies in the public interest.

Others say closer ties to President Donald Trump could give the division greater prominence and perhaps more funding.

Leadership turnover and public output

CAISSI is an independent division of the Department of Commerce and until recently was called the US Center for AI Standards and Innovation. It is allocated $10 million a year, about an eighth of the annual budget of its UK counterpart.

Its director was appointed in April 2026 and fired within days. His replacement resigned after a few months, leaving the center without a permanent leader; the head of the parent institution also serves as interim director.

Congress began asking in August 2026 why the center had stopped publishing reports on the operational safety of certain artificial intelligence systems. OpenAI also urged politicians to provide the division with more funding and support.

NIST’s official CAISSI page provides a dated log of different types of output. Its most recent listed research blog post is dated March 23, 2026, while its news section lists a September 17, 2026 assessment of Z.ai’s GLM-5.3 cyber capabilities and an October 16, 2026 publication on incident preparedness and response.

The existing voluntary framework

NIST describes CAISSI as the industry’s primary government contact for testing and collaborative research involving commercial super-intelligence systems. Its mandate includes developing measurement guidelines, best practices and voluntary standards, establishing voluntary agreements with developers and assessors, and conducting unclassified assessments of demonstrable national security risks involving cybersecurity, biosecurity and chemical weapons.

The center is also tasked with evaluating adversary systems to identify vulnerabilities, backdoors and other hidden malicious behavior. Its mandate includes coordinating evaluation methods and assessments with the Departments of Defense and Energy, the Department of Homeland Security, the Office of Science and Technology Policy, and the Intelligence Community.

Executive Order 14409, issued June 2, 2026, directs national security agencies, in consultation with NIST and others, to develop a classified process for benchmarking advanced cyber capabilities. This process is intended to establish the threshold for the designation of a “covered border model”.

The order also calls for a voluntary framework under which developers could provide the government with access to covered frontier models up to 30 days before their scheduled release to other trusted partners. We expressly state that we do not authorize mandatory government licenses, prior approvals, or permits for the development, publication, release, or distribution of artificial intelligence models.

The independence dispute is therefore about who directs and communicates technical results within that voluntary system, rather than a new federal power to approve models before release.

Company-controlled ratings

In September 2026, Trump asked major AI companies to undertake broad self-policing through a voluntary security pact. The administration has not explained how it will monitor implementation.

The evaluation problem is also changing as models are deployed as AI agents that can plan and execute multi-step tasks without ongoing human prompting. Sayash Kapoor, an incoming assistant professor at the University of California Berkeley School of Information, said there is no standard process for testing how safely agents act in real-world settings. He used the example of an agent booking a flight: checking only the end result can make the operation seem successful. “But unless you look at the entire series of steps an officer takes, you won’t be able to say the officer made a huge mistake,” he said.

Private assessments do not necessarily provide independent support. Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman have promoted independent assessment firms, but their companies can limit those companies’ access and set aside their assessments.

The White House did not respond to specific questions about CAISSI, but said Trump has formally established a Super Intelligence Force and commissioned a risk report from the Council of Advisors on Science and Technology.

Logan Koepke, senior project director at the civil rights and technology nonprofit Upturn, said internal assessments run by the company and external auditors hired by the company leave the role of government testing unclear.

Amos Toh, senior counsel at the Brennan Center for Justice, said that allowing labs to police their own models without clarifying the government’s role weakens independent controls over systems that can cause real-world harm, and that resulting decisions should be explained and justified publicly.

The reported request leaves a concrete question about governance: who sets CAISSI’s evaluation priorities and how visible its methods and results will be. Closer ties to the White House may alter its importance, but they alone do not address how independent government testing will work alongside voluntary developer access and company-controlled private reviews.