On May 29, 2026, the National Institute of Standards and Technology published a Federal Register notice — document number 2026-10779, filed under Docket No. NIST 2026-0034 — that does something more consequential than its bureaucratic surface suggests. It retitles the body that NIST first announced on November 2, 2023 as the 'Artificial Intelligence Safety Institute Consortium,' or AISIC, and re-christens it simply the 'NIST Artificial Intelligence Consortium.' Read the change in name against the change in mandate and you have a compact, official record of how the federal government's posture toward AI measurement has shifted in roughly two and a half years.
The original consortium was not a small thing. By NIST's own account in this notice, AISIC brought together more than 280 organizations to develop, in the agency's words, 'science-based and empirically backed guidelines and standards for artificial intelligence (AI) measurement, laying a foundation for global AI metrology.' That is an unusually large standing collaboration for a standards body, and it reflected how central NIST had become to the practical work of figuring out how to measure what AI systems do. The new notice does not dissolve that work. It reissues the invitation for organizations to join, revises the scope of the consortium's research, and folds the existing membership forward — current members 'are not required to reapply,' though they 'may be asked to sign amendments' reflecting the refocused direction.
What the notice asks of applicants
For any organization weighing whether to participate, the notice is admirably concrete about the terms of entry. Interested organizations are asked to 'describe the technical expertise and products, data, and/or models that they will bring to the Consortium to support the Consortium's collaborative research activities.' Participation is open to all who can contribute expertise, products, data, or models. The legal vehicle is a Cooperative Research and Development Agreement — a CRADA — which selected participants will be required to enter with NIST. The notice carves out a fallback: entities that are not legally permitted to enter CRADAs may, at NIST's discretion, participate under a separate non-CRADA agreement. That last clause matters for universities, certain nonprofits, and other organizations whose enabling statutes complicate standard CRADA terms; the agency has built a door for them rather than leaving them out.
The CRADA structure is the part that turns this from a press-release coalition into a real research instrument. A CRADA is a formal collaboration agreement that governs intellectual property, contributed resources, and the sharing of results. Requiring one means participants are not merely lending their names; they are committing data, models, or technical staff to joint work whose outputs feed directly into the measurement science NIST produces. For the AI industry, that is the most important operational fact in the notice: the metrics and methods that emerge from this consortium are the ones that other agencies — the FDA, export-control authorities, procurement offices — increasingly cite as the baseline for 'trustworthy' AI.
The directives the notice is built on
The notice is explicit about whose instructions it carries out, and reading that chain of authority is how you understand the rename. NIST cites three sources. First, its standing statutory mandate under the National Artificial Intelligence Initiative Act of 2020 (Public Law 116-283). Second, Executive Order 14179, issued January 23, 2025, titled 'Removing Barriers to American Leadership in Artificial Intelligence.' Third, 'America's AI Action Plan,' issued in July 2025. The notice states that NIST will use the consortium to 'empower the collaborative establishment of a new measurement science that will enable the identification of proven, scalable, and interoperable techniques and metrics to promote the development and use of AI.'
Set the new framing against the old name and the trajectory is visible in plain text. The 2023 body led with 'Safety.' The 2026 body leads with 'Artificial Intelligence' and frames its purpose around measurement science that promotes 'the development and use of AI.' The technical core — metrology, interoperable metrics, empirically grounded methods — is preserved and arguably strengthened. The emphasis in the surrounding language has moved from safety-as-headline toward measurement-in-service-of-development-and-leadership. This is not a contradiction; rigorous measurement is what makes both safety and innovation tractable. But the notice is a primary-source record of the executive branch's chosen emphasis, and it deserves to be read as such rather than through anyone's paraphrase.
Why it matters for the AI industry
The practical stakes for companies are real. NIST's measurement frameworks have a way of becoming the de facto standard the rest of the government leans on. The agency's AI Risk Management Framework already shows up by name in other dockets — the FDA's AI-enabled clinical-trials pilot states that it will be 'guided by principles aligned with' that framework. When NIST refocuses the research scope of a 280-plus-member consortium and reopens the membership process, it is effectively redrawing the table where the next generation of those benchmarks gets built. The organizations that sign CRADAs help shape which 'proven, scalable, and interoperable techniques and metrics' get blessed; the organizations that sit out inherit whatever the participants produce.
For the largest labs and compute providers, there is a strategic case for being inside the tent: the metrics that define 'trustworthy' AI in federal procurement and downstream regulation are negotiated, in part, through exactly this kind of collaboration. For smaller developers, the notice is a reminder that the measurement standards they will eventually be judged against are being written now, in a forum that is — on paper — open to anyone who can contribute. The rename is the headline, but the substance is in the CRADA requirement and the reopened door. As of May 29, 2026, the venue where American AI metrology is set has a new name, a revised scope, and a fresh call for members — and it remains, by NIST's own framing, the foundation for global AI measurement.