Key takeaways
- NIST released the Cyber AI Profile, NIST IR 8596, as a preliminary draft on December 16, 2025, and public comments closed January 30, 2026.
- As of October 24, 2026, NIST has not posted the initial public draft it planned for 2026, so the profile is voluntary and its wording may still change.
- It organizes AI considerations under CSF 2.0 Functions, Categories and Subcategories across three focus areas: Secure, Defend and Thwart.
- Each Subcategory gets a proposed priority of High, Moderate or Foundational for each focus area, and NIST says Foundational does not mean low priority.
- Its Govern guidance calls for a human assigned responsibility for each AI system's actions and AI-specific due diligence before adopting a third-party AI tool.
If your staff use AI notetakers, research assistants or Copilot, and your security program is built on NIST CSF 2.0, you already face the question of where AI fits in that program. A client due diligence questionnaire, an examiner or your board may ask it before NIST finishes its own answer.
This guide explains what the NIST Cyber AI Profile says, how current it is, and what a CCO, COO or managing partner can do with it this quarter.
What is the NIST Cyber AI Profile?
The NIST Cyber AI Profile is NIST's draft guide to managing AI-related cybersecurity risk using the Cybersecurity Framework 2.0. NIST IR 8596, the Cybersecurity Framework Profile for Artificial Intelligence, was released as a preliminary draft on December 16, 2025. NIST describes it as guidelines for managing cybersecurity risk related to AI systems and for identifying opportunities to use AI to strengthen cybersecurity, organized using the CSF 2.0 Functions, Categories and Subcategories.
For a chief compliance officer, COO or managing partner, that structure is the useful part. If your security program is already organized around CSF 2.0, the profile shows where AI fits inside it, so AI risk appears in the same profiles, vendor reviews and reports you already produce.
Is the NIST Cyber AI Profile final or mandatory?
The NIST Cyber AI Profile is neither final nor mandatory. The CSF itself is voluntary guidance, and the profile is still a preliminary draft. Public comments closed January 30, 2026, and NIST said it planned to develop an initial public draft for release in 2026. As of October 24, 2026, NIST's publication page still lists only the December 16, 2025 preliminary draft, and the NCCoE project page shows its status as reviewing comments. NIST held virtual working sessions on updates to the profile on April 28, May 5 and May 12, 2026. Expect the priorities and wording to change before a final version.
Reading it now still pays. It shows the structure NIST is building for AI, and a firm that organizes its AI controls that way today can adjust to the next draft with little rework.
What are the three focus areas of the Cyber AI Profile?
The Cyber AI Profile's three focus areas are Secure, Defend and Thwart, and each one is scored against the CSF 2.0 Subcategories. The preliminary draft defines them this way:
- Secure (Securing AI System Components): managing cybersecurity challenges when integrating AI into the organization's systems and infrastructure. For most firms, this is the AI tools staff already use.
- Defend (Conducting AI-Enabled Cyber Defense): identifying opportunities to use AI to improve cybersecurity processes and activities.
- Thwart (Thwarting AI-Enabled Cyber Attacks): building resilience to protect against new AI-enabled threat vectors.
Each Subcategory receives a proposed priority for each focus area: 1 (High), 2 (Moderate) or 3 (Foundational). NIST is explicit that Foundational does not mean low priority and that all Subcategories should receive consideration. It also notes that priorities may be higher or lower for individual organizations, which leaves room for a small firm to set its own order.
What does the Cyber AI Profile say about governing AI?
The Cyber AI Profile's Govern guidance says a named person should be responsible for each AI system's actions and that AI suppliers need more scrutiny than ordinary software vendors. The Govern function is where the draft is most concrete, and where a leadership team can act without technical work:
- Ownership (GV.RR-02). The draft says a human is assigned responsibility for the actions of an AI system. An AI notetaker or research assistant should have a named owner, the same way your CRM or document management system does.
- Supply chain (GV.SC-01 and GV.SC-06). With AI, the draft says data provenance should be weighted as heavily as software and hardware origin. It also suggests AI-specific due diligence and adversarial testing before adopting a third-party AI solution, and confirming that a supplier's training methods and input data are accessible and transparent.
- Incident response (GV.SC-08). Suppliers that provide AI models or data should be integrated into AI-specific incident response and recovery planning.
- Policy and risk tolerance (GV.PO-02 and GV.RM-02). Policies that affect AI use may need more frequent updates, and new risk tolerance statements may be needed for AI-enabled threats.
None of this requires a new framework. It requires adding AI to the inventory, vendor reviews and policies that a CSF-based program already maintains.
How is the Cyber AI Profile different from the NIST AI RMF?
The Cyber AI Profile covers the cybersecurity risks of AI in CSF 2.0 terms, while the NIST AI Risk Management Framework (AI RMF) addresses responsible AI use more broadly. NIST says the profile works alongside existing frameworks and will map to other NIST resources, including the AI RMF. The AI RMF's Generative AI Profile (NIST AI 600-1) identifies twelve risks unique to or made worse by generative AI, including data privacy and information security. NIST is also developing SP 800-53 control overlays for securing AI systems, which will be more implementation-focused.
For a smaller firm, a workable division is simple: use the CSF 2.0 and the Cyber AI Profile to run and report the security program, and use the AI RMF material when the board or a client asks broader questions about how AI is used.
What should your firm do about the Cyber AI Profile this quarter?
Your firm should start with the Secure focus area and the Govern function this quarter, because that is where most firms already have AI exposure:
- List every AI tool in use, including AI features switched on inside software you already license, and note what data each one touches.
- Assign a named owner to each AI system, and record that assignment where your other system owners are recorded.
- Add AI suppliers to your vendor review with questions on data retention, model training and incident notification. Our guide to vetting AI vendor data terms lists what to ask.
- Update your CSF Current and Target Profiles so the AI-related Subcategories show up in the gap list and the plan.
- Report AI progress to leadership in the same terms as the rest of the program. Our guide to CSF 2.0 metrics by function covers what to track.
- Set a reminder to revisit the work when NIST publishes the initial public draft.
This article summarizes what the draft says. How your regulators expect AI risk to be handled in your specific situation is a question for your compliance counsel.
How can s90 help you apply the Cyber AI Profile?
s90 helps compliance-focused firms in Los Angeles and Orange County fold AI into a NIST CSF 2.0 program. We use CSF 2.0 as the common language across the regulations our clients answer to, and we help firms find the AI tools already in use, put guardrails and alerts around new ones, and add AI to the profiles, vendor reviews and reports leadership already sees. If you want a clear picture of where AI sits in your program, start with a conversation.
