Featured in Citywire & AP News — Book a 15-minute intro
    AI exam readiness 2026 for RIAs
    Back to Insights
    Exam Preparation

    AI Exam Readiness 2026: How RIAs Should Prepare for AI-Focused SEC Reviews

    July 8, 2026
    •
    11 min read

    SEC examiners have moved past general questions about whether RIAs use artificial intelligence. In 2026 they are asking pointed operational questions: which models, whose data, what testing, what disclosures, and what happens when the model gets it wrong. Firms should assume AI will be a named topic in the next exam and prepare accordingly.

    What Examiners Are Actually Asking

    Recent request lists show examiners asking for a written inventory of AI tools in use, the name of the business owner for each, sample outputs, testing records, vendor contracts, disclosures made to clients, and evidence that compliance reviewed the tool before launch. Firms that cannot produce those artifacts quickly end up on a slower, more intrusive exam track.

    Model Inventory

    A single source of truth listing every AI tool, its purpose, its owner, and its risk tier.

    Human Oversight

    Evidence that a qualified person reviewed AI outputs before they reached clients.

    Disclosure Match

    Form ADV, brochures, and marketing pages consistent with real AI usage.

    Incident Trail

    Logged instances where AI output was corrected, escalated, or rolled back.

    Build the AI Evidence File Now

    The strongest firms are assembling a dedicated AI evidence file that mirrors the way exam requests arrive. It contains the model inventory, testing workpapers, vendor due diligence, sample outputs with reviewer initials, training records, disclosure change logs, and a short narrative that explains the firm's AI governance in one page. That narrative gives examiners a place to start and helps the CCO tell a consistent story.

    Model Risk and Human-in-the-Loop

    Not every AI use case carries the same risk. A meeting summarizer used for internal notes is different from an assistant that drafts client communications or influences allocation decisions. Firms should tier their tools, apply proportionate controls, and keep evidence that higher-risk uses receive stricter review. Human-in-the-loop should be a real step with a real reviewer, not a checkbox at the end of an automated workflow.

    Marketing and Disclosure of AI Use

    Marketing Rule reviewers are paying attention to AI claims. Statements about personalized advice, proprietary intelligence, or automated insights should be supported by documentation. If AI generates or shapes public content, disclosures should describe the role of AI, and archived materials should reflect the same standards as current ones.

    Tabletop Preparation Checklist

    1. Run a mock exam request focused only on AI and see what your firm can produce in 48 hours.
    2. Confirm the model inventory matches what employees actually use, including free tools.
    3. Test that human review steps have named reviewers and dated evidence.
    4. Review vendor files for AI-specific due diligence and updated contracts.
    5. Reconcile Form ADV, brochure, and website language with current AI practice.
    6. Rehearse the one-page AI governance narrative with the CCO and business owners.

    Preparing for an AI-focused SEC exam?

    NextReg runs AI-focused mock exams, builds evidence files, and helps firms tighten model governance before regulators arrive.

    Schedule a Consultation