The Legal Stack
Independent LegalTech Analysis
← Research Briefings
Research BriefingNo. 116 · November 20, 2026 · 10 min read
Compliance & Regulation · Research Report

What Published SEC Enforcement Actions, Staff Guidance, and No-Action Letters Through October 2026 Actually Say About AI-Assisted Legal Opinions, Disclosure Drafting, and the Attorney Liability Exposure When AI-Generated Securities Filings Contain Material Misstatements

The Core Regulatory Gap

The SEC has not issued a formal rule, no-action letter, or enforcement release that directly addresses what happens when a registration statement, proxy statement, or periodic report contains a material misstatement or omission that originated in or was perpetuated by an AI-assisted drafting tool. That gap is not an oversight — it reflects the agency's consistent posture of applying existing frameworks to new technologies rather than issuing technology-specific safe harbors. The practical consequence for outside securities counsel is that the liability analysis runs entirely through statutory and regulatory provisions that predate generative AI, applied to fact patterns that those frameworks were not designed to contemplate.


What the SEC Has Actually Said About AI in Disclosure

The most substantive SEC guidance addressing AI in the disclosure context came through a wave of Commission-level speeches, comment letter campaigns, and regulatory proposals in 2023–2024, alongside the Commission's focus on what the agency itself labeled "AI washing."

Corp Fin comment letter campaigns and signaling (2023–2024). The Division of Corporation Finance did not issue a dedicated AI disclosure rule before October 2026, but its 2023 comment letter campaigns directed at registrants in technology-adjacent sectors made clear that the staff expected companies to describe their actual use of AI in operations with specificity sufficient to avoid generic boilerplate. Comment letters — publicly available on EDGAR — repeatedly pushed back on vague risk-factor language that acknowledged AI use in one breath and disclaimed all material reliance in the next. While these letters are staff-level positions rather than binding law, they signal where materiality analysis sits.

The "AI washing" enforcement actions (March 2024). The Commission took its most concrete AI-related enforcement steps in March 2024, bringing settled administrative proceedings against two registered investment advisers — Delphia (USA) Inc. and Global Predictions Inc. — for making materially false and misleading statements about their use of AI and machine learning in investment processes. Both orders were brought under the Investment Advisers Act, not the Securities Act, but the underlying theory — that overstating or mischaracterizing AI capabilities in investor-facing communications constitutes a material misrepresentation — translates directly to the disclosure context. The orders did not implicate outside counsel's drafting role, but they establish the Commission's willingness to treat AI-related mischaracterizations as enforcement-grade misstatements.

No-action letters. Through October 2026, no published no-action letter addresses AI tool use in the preparation of opinion letters, registration statements, or disclosure documents. The no-action process has not been used by outside counsel to seek comfort on AI-assisted drafting workflows, and the staff has issued no guidance indicating it would respond favorably to such requests.


How Section 11 and Rule 10b-5 Apply When AI Is in the Drafting Chain

The existing liability framework for outside securities counsel rests on two pillars, neither of which contains a carve-out for AI-originated content.

Section 11 of the Securities Act. Section 11 imposes liability for material misstatements or omissions in registration statements on every person who signed the registration statement and on experts who consented to the use of their expert reports within it. Outside counsel typically avoids Section 11 "expert" status by declining to consent to inclusion of legal opinions as expertised portions of the registration statement. However, counsel who do consent — most commonly in connection with tax opinions or enforceability opinions incorporated by reference — accept Section 11 strict liability for the portions they expert, with a due diligence defense available only for non-expertised portions. If an AI tool generated or materially shaped the analysis underlying an opinion that was then incorporated as an expertised portion, the attorney's ability to demonstrate the kind of reasonable investigation required for the due diligence defense becomes a live factual question. The standard is objective: what would a prudent person in that position have done? Courts and the Commission have not yet addressed whether relying on unverified AI output in forming an expert opinion satisfies that standard, but the question is not abstract — it tracks the same competence logic that the ABA applied in Formal Opinion 512, issued July 29, 2024, which addressed lawyers' obligations to independently verify AI-generated output.

Rule 10b-5. Primary liability under Rule 10b-5 for a misstatement in a securities filing requires a false statement of material fact, scienter, and reliance. Outside counsel drafting on behalf of an issuer client faces aiding-and-abetting exposure under Exchange Act Section 20(e) rather than primary liability in most scenarios, because the attorney typically does not make the statement — the issuer does. The scienter element for aiding and abetting requires knowledge of the fraud or reckless disregard of a substantial risk of fraud. If counsel uses an AI tool that generates factually incorrect disclosure language — say, a mischaracterization of a pending regulatory proceeding — and counsel fails to verify that output before it goes into a filing, the question becomes whether that failure to verify constitutes reckless disregard. The SEC has not articulated a specific standard, but its broader enforcement posture on professional responsibility suggests that over-reliance on automated output without independent verification is precisely the kind of conduct that could support a recklessness finding.


What the SEC's Own AI Disclosures Signal About Expectations

The SEC's Office of Inspector General and the Division of Examinations have each disclosed the agency's piloting of AI-assisted examination tools in public reports and testimony available on the SEC's own website. The Division of Examinations' examination priorities have noted the use of data analytics and technology-assisted review in evaluating registrant filings. While these agency-side disclosures do not impose any direct obligation on outside counsel, they carry an implicit signal: the Commission regards AI-assisted review as a legitimate workflow when accompanied by human verification and oversight. The agency is not telling regulated parties to avoid AI — it uses AI itself. What it is signaling, through both enforcement posture and its own governance practices, is that human accountability for the output is non-negotiable.


How Outside Securities Counsel Should Be Documenting AI Tool Use Now

In the absence of a safe harbor, documentation serves both a defensive and a due diligence function. Based on the frameworks that do exist — Section 11's due diligence standard, the ABA's competence and candor obligations under Formal Opinion 512, and the SEC's own track record of scrutinizing process failures in enforcement — the following documentation practices have clear analytical grounding.

Engagement-level disclosure to the issuer client. Counsel should ensure the engagement letter or a written addendum describes whether AI tools will be used in the drafting or review of disclosure documents, identifies the categories of tools, and specifies that the attorney of record retains full responsibility for all output. This creates a clear record that the client understood and assented to the workflow.

Version control and audit trail. For any AI-generated draft of disclosure language subsequently incorporated into a filing, counsel should retain a contemporaneous record showing what AI-generated language was reviewed, what was changed, and by whom, before the language was submitted to the issuer for approval. This is not a legal requirement under any current SEC rule, but it is the kind of documentary record that supports a due diligence defense.

Verification protocols for factual assertions. The SEC's comment letter practice on AI disclosure language, combined with the AI washing enforcement actions, makes clear that factual assertions about AI — and, by extension, any factual assertion produced by AI — require independent verification traceable to primary sources. Counsel should build explicit verification steps into drafting workflows rather than treating AI output as a starting point requiring only light editing.

Issuer-side AI governance as a disclosure consideration. Outside counsel reviewing draft periodic reports or registration statements should flag whether the issuer has adequately disclosed its own use of AI in operations, risk factors, and internal controls — particularly where AI is used in financial reporting processes — given the ongoing staff comment letter focus on that area.


Bottom Line

The SEC's published record through October 2026 leaves the liability framework for AI-assisted disclosure drafting entirely within the architecture of Section 11 and Rule 10b-5 as written. The agency has demonstrated, through the Delphia and Global Predictions actions, that it treats AI mischaracterizations as enforcement-grade misstatements. It has not created any AI-specific safe harbor, and no no-action letter provides comfort for AI-assisted opinion work. Outside securities counsel's exposure is therefore governed by the same due diligence, verification, and professional responsibility standards that have always applied — with the unresolved factual question being whether, and under what circumstances, reliance on AI output without sufficient independent verification will satisfy those standards when tested in an enforcement or litigation context.

Filed under Compliance & Regulation → · The Legal Stack accepts no vendor funding for its research.

More Research

View all →
No. 115
10 min
What Published IRS Guidance, Tax Court Opinions, and Treasury Proposed Rulemaking Through October 2026 Actually Say About AI-Assisted Tax Legal Work Product — Privilege, Disclosure Obligations, and the Penalty Exposure When AI-Generated Positions Turn Out to Be Wrong
10 min
No. 114
10 min
What Published FTC Guidance, State AG Enforcement Actions, and Bar Ethics Opinions Through October 2026 Actually Say About AI-Generated Legal Intake Tools That Collect Consumer Data Before Any Attorney-Client Relationship Is Formed
10 min
No. 113
10 min
What Published USPTO Guidance, Federal Circuit Opinions, and PTAB Decisions Through October 2026 Actually Say About AI-Assisted Patent Prosecution, Inventorship Disclosure Obligations, and the Duty of Candor When AI Tools Generate Claim Language
10 min
© 2026 The Legal Stack — Independent LegalTech Analysis