Officials at the US government’s AI safety center were recently asked to “informally report to an office of the White House”, the Associated Press reported on 8 November, citing two former government officials who spoke on condition of anonymity. The center is the Commerce Department’s Center for Advancing Innovation and Standards for Super Intelligence, or CAISSI.
The AP says the center “is allocated $10 million per year”, about one-eighth of its UK counterpart’s annual budget, lacks a permanent leader and has stopped publishing its research. NIST’s own pages list four model assessments by the center in 2026; by our arithmetic, the latest, dated 17 September, came 52 days before the AP report.
Business Pill · Guardrails
A short Business Engineer explainer on AI guardrails. It teaches the general idea only and does not discuss CAISSI or the AP report.
The key insight: The Associated Press reports, citing two former government officials, that staff at the Commerce Department’s AI safety center, CAISSI, were recently asked to informally report to an office of the White House. The AP says the center is allocated $10 million a year and has no permanent leader; NIST’s pages list four model assessments this year, the latest on 17 September, 52 days before the report by our arithmetic.
The AP does not name the White House office. It says the arrangement “concerns some critics”, who worry the executive branch might be politicizing decisions on risks from cybersecurity to bioweapons, and that the center’s findings “could be altered or muted in deference to the tech industry”.
The AP also reports that others argue bringing the officials closer to President Donald Trump “could give the division more prominence — and perhaps more funding”.
On leadership, the AP says the division’s director was hired and then fired within days in April, the replacement resigned after a few months, and the head of its parent institute doubles as acting director. The center was called the U.S. Center for AI Standards and Innovation until recently.
In August, the AP says, Congress started asking why the center stopped making public its reports on how safely certain AI systems were performing. OpenAI had by then asked policymakers to put more money and support behind the division, per the AP.
Model assessments listed on NIST’s CAISSI page in 2026 by publication date, against the Associated Press report of 8 November. The 52-day gap is FourWeekMBA’s arithmetic on NIST’s dates.
What the White House Said
The White House did not respond to specific questions about CAISSI, the AP says. In a statement, it said Trump had formally established his new Super Intelligence Force, has commissioned a report on risks from his Council of Advisors on Science and Technology, and will ensure “continued U.S. leadership on SI for the benefit of the American people”.
What the Center Does
NIST’s CAISSI page says the center will serve as industry’s primary point of contact within the U.S. government for testing and research on commercial SI systems, and will “establish voluntary agreements with private sector SI developers and evaluators”.
In May, NIST announced agreements with Google DeepMind, Microsoft and xAI for pre-deployment evaluations, building on OpenAI’s and Anthropic’s agreements from 2024, The Hill reported. The Hill says the office had completed more than 40 evaluations of this nature.
What NIST Has Published in 2026
The CAISSI page lists four model assessments in 2026: DeepSeek V4 Pro on 1 May, Z.ai’s GLM-5.2 on 17 July, a joint preliminary assessment of Kimi K3’s cyber capabilities with the UK AI Security Institute on 23 July, and Z.ai’s GLM-5.3 on 17 September. None of the four is of a model from the five labs with testing agreements.
On 16 October, the center published incident-response practices with the NSA’s Super Intelligence Security Center and CISA. From the 17 September assessment to the AP report on 8 November is 52 days, by our arithmetic.
What Critics and Researchers Told the AP
Caroline Siegel Singh of the Federation of American Scientists told the AP: “These are really powerful institutions that can drastically improve our lives, or the alternative”.
Logan Koepke of Upturn asked: “Like, what role does government testing actually play in that scenario?” Amos Toh of the Brennan Center told the AP the decisions “should be explained and rationalized in the public eye”.
As we read the AP report, the change it describes is a reporting line, not a new law or order: the AP cites no written directive and does not name the White House office.
As we read NIST’s pages, the center’s public model assessments this year cover DeepSeek, Z.ai and Kimi models, while the five labs with testing agreements are US companies.
As we read the White House statement, it answers with the Super Intelligence Force and a PCAST risk report rather than with the center’s status.
Caroline Siegel Singh, Federation of American Scientists, to the Associated Press, 8 November 2026
“These are really powerful institutions that can drastically improve our lives, or the alternative.”
Three Implications
FOR AI LABS NIST says the center is industry’s primary point of contact for testing; Google DeepMind, Microsoft, xAI, OpenAI and Anthropic have agreements with it, per NIST and The Hill.
FOR OVERSIGHT The AP says Congress began asking in August why the center stopped making its safety reports public.
FOR THE BUDGET The AP puts the center’s allocation at $10 million a year, about one-eighth of the UK counterpart’s.
The analysis says: “The governance layer sits at the intersection of capability, regulation, national security, access control, and geopolitical alignment.”
As we read it, the AP report is about who sits closest to that layer inside the US government: a testing center at Commerce, or an office of the White House. The analysis predates the report and does not discuss CAISSI.
What Is Not Established
The AP’s account rests on two former officials speaking anonymously; it does not name the White House office or describe the reporting line in writing. The AP piece carries no on-record comment from the Commerce Department, NIST or the AI labs. We did not contact them. The 52-day figure is our arithmetic on NIST’s dates.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
The AP reports that staff at the $10 million-a-year US AI safety center were asked to informally report to a White House office, citing two former officials. NIST’s pages list four model assessments by the center this year, the latest on 17 September; the White House did not answer specific questions about the center.
A note on sourcing. The account of the reporting arrangement is the Associated Press’s, citing two former government officials, read in full via PBS NewsHour on 9 November 2026. Assessment dates and the center’s remit are from NIST’s pages, read the same day; the May testing agreements are from a NIST bulletin and The Hill. We did not contact the White House, the Commerce Department or NIST. The 52-day figure is our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
