meilynx

Reference · Securities

SEC FY2027 Examination Priorities: What Examiners Will Test on AI

What the Division of Examinations will review about AI in fiscal year 2027, and what changed from fiscal year 2026

Last reviewed November 15, 2026

At a glance

Published
23 October 2026
Issued by
SEC Division of Examinations (press release 2026-109)
Legal effect
A staff statement. It is not a rule and creates no new obligations.
Term for AI
Super intelligence (SI)
Where AI appears
Emerging Financial Technology (p. 12), Regulation S-ID and Regulation SCI (p. 13)

The SEC Division of Examinations published its priorities for fiscal year 2027 on 23 October 2026. Emerging technologies are among the key areas of focus the Commission announced with them.SEC release 2026-109 The document tells broker-dealers, investment advisers and other registrants where examiners will look in the year ahead. It is a staff statement: it is not a rule, the Commission has neither approved nor disapproved it, and it creates no new obligations for anyone.SEC FY2027 Priorities · Disclaimer

The document never uses the words artificial intelligence. It uses the term "super intelligence (SI)" where the fiscal year 2026 priorities said AI, and gives no definition.SEC FY2027 Priorities · p. 12SEC FY2026 Priorities · p. 13 Its examples are ordinary generative tools: registrants use generative SI tools for research, drafting, coding and client communications, for advisory and portfolio management, and for compliance monitoring and surveillance.SEC FY2027 Priorities · p. 2 This page says AI, with the SEC's term in parentheses where the wording matters.

The Division says it is working to understand how these technologies influence outcomes and whether firms continue to comply with "existing, technology-neutral obligations."SEC FY2027 Priorities · p. 2

Examiners will test whether automated advice matches what investors were told

The Emerging Financial Technology topic covers automated investment tools, trading algorithms or platforms, and the risks of emerging technologies and alternative data sources. Firms that offer automated investment advisory services, recommendations and related tools are the Division's priority.SEC FY2027 Priorities · p. 12

Reviews of these tools will generally assess three things: whether representations are fair and accurate, whether operations and controls are consistent with disclosures made to investors, and whether algorithms lead to advice or recommendations consistent with investors' investment profiles or stated strategies.SEC FY2027 Priorities · p. 12

Each of the three sets a statement against what a system did. A description of an AI capability in marketing material or a disclosure is checked against the tool it describes, and a recommendation against the profile it was meant to fit.

Policies must cover AI used in fraud detection, back-office work, AML, Reg BI and trading

The Division will prioritize examinations of policies and procedures for monitoring AI (SI), including its use in fraud prevention and detection, back-office operations, compliance with anti-money laundering and Regulation Best Interest obligations, and trading functions.SEC FY2027 Priorities · p. 12

The document describes registrants using SI in core business operations and systems, including investment analysis, customer management and code development.SEC FY2027 Priorities · p. 12 The review starts with which AI tools are in use, in which of those functions, and who supervises each.

AI-enabled attacks are now an exam topic

Threat actors are using SI to automate and accelerate cyberattacks, making them more efficient and harder to detect. Examinations will focus on compliance with the rules on how registrants assess and prepare for SI-related cyberattacks that may affect their businesses or customers. The Division's examples are SI-generated phishing messages and synthetic or deepfake voice and video used for social engineering.SEC FY2027 Priorities · p. 12

Under Regulation S-ID, examiners will review how firms identify, detect and respond to red flags of identity theft, particularly during customer account takeovers and fraudulent transfers. They will assess whether firms are updating their Identity Theft Prevention Program to address emerging risks "such as those related to SI and stolen credentials."SEC FY2027 Priorities · p. 13

The review asks whether the program is being updated: an examiner reads the date of its last update and the red flags that update added.

Reg SCI entities must account for AI and their reliance on vendors

Examinations under Regulation SCI will focus on emerging trends, and the Division names SI alongside tokenization of securities, extended trading hours and quantum computing. It will also review SCI entities' reliance on cloud and other third-party vendors, particularly how they manage vendor risk and identify vendor systems that qualify as SCI systems or indirect SCI systems.SEC FY2027 Priorities · p. 13

A model an SCI entity reaches through a vendor's service is a vendor system like any other, and the classification question applies to it.

FY2027 renames the technology and widens where it is examined

The fiscal year 2026 priorities, published 17 November 2025, already made AI an examination topic. The Division said it would review registrants' representations about their AI capabilities for accuracy, and assess policies and procedures to monitor or supervise the use of AI for fraud prevention and detection, back-office operations, anti-money laundering and trading.SEC FY2026 Priorities · p. 13

In 2026 the cybersecurity review already covered new risks associated with AI and polymorphic malware attacks.SEC FY2026 Priorities · p. 12 For fiscal year 2027 the Division adds Regulation Best Interest to the list of uses, names SI-generated phishing and deepfakes as the attacks to prepare for, and brings AI into its Regulation S-ID and Regulation SCI reviews.SEC FY2027 Priorities · p. 12SEC FY2027 Priorities · p. 13 Two sentences from 2026 are not repeated: the separate review of representations about AI capabilities, and a fourth test on controls confirming that automated advice meets regulatory obligations to investors, including retail and older investors.SEC FY2026 Priorities · p. 13 The general test that representations are fair and accurate stays.

A topic left out of the 2027 priorities remains open to an examiner.SEC FY2027 Priorities · p. 5

Before and after

AI in the SEC examination priorities, fiscal year 2026 and 2027

FY2026 (17 November 2025)FY2027 (23 October 2026)
Term for the technology
Artificial intelligence (AI)
Super intelligence (SI)
Uses the monitoring policies must cover
Fraud, back office, AML, trading
Adds Reg BI compliance
Review of AI capability claims
Named on its own
Not restated
Tests for automated advice
Four
Three
AI-enabled attacks
AI and polymorphic malware
SI phishing, deepfake voice and video
Regulation S-ID
Red flags and firm training
Program updated for SI and stolen credentials
Regulation SCI
Incident response and vendor risk
SI as an emerging trend, cloud and vendor reliance
Regulatory technology
Integration reviewed
Not restated
From the FY2026 Examination Priorities, pages 12 and 13, and the FY2027 Examination Priorities, pages 12 and 13. "Not restated" means the 2027 text does not repeat the 2026 sentence.

The adviser and broker-dealer sections do not name AI

The Compliance Programs section for investment advisers and the Supervision section for broker-dealers do not mention AI. Both name change as a trigger. Examiners will focus on advisers that have changed their business models or started providing new services, and assess whether the compliance program was adapted in response.SEC FY2027 Priorities · p. 6 Broker-dealer examinations will evaluate whether supervisory systems, written policies and procedures, and internal controls are reasonably designed, including those for new business activities and investment products, and whether qualified individuals hold supervisory authority.SEC FY2027 Priorities · p. 9

The document does not say whether adopting an AI tool counts as a new service or a new business activity.

Evidence an examiner can request

The priorities list no document requests. Each review above is still answered with records a registrant keeps or can produce:

  • An inventory of the AI tools in use, the business function each serves (fraud detection, back office, AML, Reg BI, trading, client communications) and who approved it.
  • The written policies and procedures for monitoring each tool, and records showing the monitoring ran.
  • For each statement to investors about an AI capability, the tool it describes and records of how that tool operates.
  • For automated advice, records tying each recommendation to the investor profile or stated strategy it was checked against.
  • The current Identity Theft Prevention Program, its last update date and the red flags it covers for synthetic voice, video and stolen credentials.
  • For SCI entities, the vendor inventory and the classification of each vendor system, AI services included.
  • The books and records Rule 17a-4 and Rule 204-2 already require, AI-drafted communications included. The SEC recordkeeping entry covers which AI records those rules reach, and Does generative AI client outreach trigger SEC and FINRA recordkeeping rules? works through one common case.
FINRA member firms face a parallel set of expectations in the FINRA 2026 Oversight Report. The questions examiners open with in conversation are collected in What examiners are asking about AI in 2026.

Control mapping

What a reviewer expects to be able to see.

ObligationWhat the system must doEvidence a reviewer expects
Representations (Emerging Financial Technology)Keep statements about AI and automated tools fair and accurateEach AI claim matched to the tool it describes and records of how that tool runs
Operations and disclosuresRun automated tools and their controls as disclosed to investorsDisclosures kept alongside the configuration and operating records of the tools they describe
Algorithmic adviceKeep automated advice consistent with investors' profiles or stated strategiesRecords tying each recommendation to the profile or strategy it was checked against
Monitoring AI useMonitor AI used in fraud detection, back office, AML, Reg BI and tradingAn AI tool inventory by function, the written procedures, and monitoring records
AI-enabled cyberattacksAssess and prepare for SI-generated phishing and deepfake social engineeringA dated risk assessment and the controls it led to
Regulation S-IDUpdate the Identity Theft Prevention Program for SI and stolen credentialsThe dated program and the red flags it now covers
Regulation SCIAccount for SI and for reliance on cloud and third-party vendorsA vendor inventory with the SCI or indirect SCI classification of each system
Books and records (17a-4, 204-2)Keep AI-generated records the existing rules requireRetained records, retrievable in the required form

Key dates

  • 17 November 2025SEC Division of Examinations publishes its fiscal year 2026 priorities, naming AI under Emerging Financial Technology.
  • 23 October 2026SEC Division of Examinations publishes its fiscal year 2027 priorities (press release 2026-109), using the term super intelligence (SI).

Sources cited

Common gaps

Where a registrant's AI program usually falls short of what these reviews test.

  • Claims ahead of the system. A brochure or pitch that says a product uses AI to select investments or watch risk, with nothing on file showing the tool does that. Representations are tested against operations.
  • An inventory built by survey. A list of AI tools compiled by asking teams misses features embedded in vendor products and tools nobody reported.
  • Reg BI left out of the AI policy. A policy written to the 2026 list covers fraud, back office, AML and trading, and says nothing about recommendations made with AI support.
  • Monitoring with no record. A procedure that says someone reviews AI output, with nothing showing who reviewed what and when, cannot be evidenced.
  • Red flags that predate synthetic voice. An Identity Theft Prevention Program last updated before cloned voices and deepfake video were common, with no red flags for either.
  • Vendor AI left unclassified. An SCI entity that has not assessed a model reached through a vendor service as a possible SCI or indirect SCI system.

Last reviewed November 15, 2026. This reference summarises publicly available regulatory guidance and is provided for general information. It is not legal advice. Obligations depend on an institution's charter, registration status, size, and activities. Verify against the primary sources cited above and consult counsel before relying on any summary here.

Regulatory updates

When a regulator changes what an AI examination asks for, hear about it first.

Short notes on SR 26-2, NYDFS 500, FINRA, the NAIC bulletin, the EU AI Act, and the employment-AI statutes, plus what we ship. A few emails a month.