NIST Cyber AI Profile (IR 8596): what the preliminary draft covers and how far to rely on it

Why this page: security questionnaires and RFPs have started to ask whether a supplier follows “the NIST Cyber AI Profile”. The site covers NIST AI 100-2 attack terms and NIST AI 600-1, but not this document. It is a draft, so this page is mostly about status. Facts below come from NIST’s publication page as fetched on 16 October 2026 and from the draft PDF’s title and front matter.

Status: a preliminary draft, comments closed

NIST’s page labels the document NIST IR 8596 (Initial Preliminary Draft) with Date Published: December 16, 2025 and Comments Due: January 30, 2026 (public comment period is CLOSED). The draft’s own abstract says This Preliminary Draft is shared along with a request for public comment to solicit feedback on the planned direction and content. and that Comments received will inform the initial public draft.

So it is two steps before a final publication: an initial public draft comes next, then a final. This page’s check on 16 October 2026 did not find an initial public draft at the NIST address we tried, so treat the content as subject to change. The PDF is at the NIST publication server.

What it says it is

The abstract says the Profile will provide guidelines for managing cybersecurity risk related to AI systems as well as identifying opportunities for using AI to enhance cybersecurity capabilities and is organized using the NIST Cybersecurity Framework 2.0 outcomes (Functions, Categories, and Subcategories).

NIST’s page lists three focus areas: Securing AI System Components (Secure), Conducting AI-Enabled Cyber Defense (Defend) and Thwarting AI-enabled Cyber Attacks (Thwart). Only the first is about securing the AI your organization builds or buys. The other two are about using AI in defense and resisting AI-enabled attacks.

The page also carries a notice that NIST is working to update its communications to incorporate the term “super intelligence” as directed. The document’s title in the PDF still says Artificial Intelligence, and this site cites it that way; see the EO 14434 note.

How it sits next to NIST’s other AI security work

NIST’s COSAiS project page says the control overlays will also leverage NIST SP 800-218A, Draft NIST AI 800-1, and NIST AI 100-2e2025. NIST’s Cyber AI Profile page announced a January 2026 workshop to discuss the Preliminary Draft of the Cyber AI Profile and to hear updates on COSAiS. The Profile is outcome-level (CSF subcategories); the overlays are control-level (SP 800-53). See COSAiS overlays.

What a supplier or buyer can do now (editorial judgment)

This section is the site’s judgment, not NIST’s.

• Do not claim conformance. A draft profile has no conformance statement, and NIST’s page does not offer one. A supplier that claims to “comply with the Cyber AI Profile” is claiming something that cannot yet be checked.
• Ask the CSF 2.0 question instead. Ask whether the supplier’s cybersecurity program is already organized on CSF 2.0 and which AI systems are in scope. That evidence will map onto the Profile when it is final.
• Use it as a checklist of topics. Secure, Defend and Thwart are a useful split for a risk register, even before the content settles.
• For CUI work, stay with the controls you are assessed on. CMMC and NIST SP 800-171 remain the assessed requirement; nothing on NIST’s page changes that.

Questions

Is the Cyber AI Profile a regulation or a certification?

NIST’s page describes a draft profile. It does not describe a regulation or certification scheme.

Where do I find the document?

NIST’s page lists the publication at https://doi.org/10.6028/NIST.IR.8596.iprd, which resolves to the PDF linked above.