Generative AI has become a drafting tool for offering memoranda, pitch decks, risk factors, and SEC comment letter responses. The output can be polished, well-organized, and persuasive. It can also be wrong, outdated, or internally inconsistent in ways that are difficult to spot. For securities lawyers and issuers, the question is not whether AI can help — it can — but which legal responsibilities remain with the human beings who sign the filing.
This article explains where AI fits into the disclosure process, where it does not, and how the SEC is likely to evaluate an issuer that relies on AI-generated content in an offering document.
What AI Does Well in Securities Work
Used carefully, generative AI can assist with:
- › First drafts of routine sections. Risk factor language, business descriptions, and definitional sections can be drafted quickly and then verified.
- › Summarizing long documents. AI can help extract themes from comment letters, contracts, or prior filings for lawyer review.
- › Consistency checks. Comparing language across a prospectus, deck, and website to identify contradictions.
- › Regulatory research assistance. Surfacing rules, releases, and enforcement actions that counsel then verifies.
These uses are valuable because they reduce mechanical work and surface issues for human judgment. They do not replace legal judgment.
What AI Cannot Do
AI models do not know the facts of a particular company. They predict plausible-sounding text based on patterns in training data. That means they cannot:
- › Verify that a representation is true.
- › Know whether a projection has a reasonable basis in the company's actual records.
- › Identify a material omission unique to the issuer.
- › Understand the current state of negotiations, product development, or regulatory discussions.
- › Apply the securities-law standard of "materiality" to a specific set of circumstances.
An issuer that files an AI-generated disclosure without human verification is making a bet that the model's training data happens to match the company's reality. That is not a defensible disclosure process.
The SEC's Likely Framework
The SEC has not issued a comprehensive rule on AI-generated disclosures, but its existing framework already applies. The antifraud provisions of the Securities Act and Exchange Act, the due diligence obligations of company officers and counsel, and the accountant's responsibilities for financial statements do not change because a model produced the first draft.
Several principles are already clear:
- › The signer is responsible. Officers who sign an S-1 or certify periodic reports are responsible for the accuracy and completeness of the disclosure, regardless of how it was drafted.
- › Material misstatements are material misstatements. The SEC will not treat an AI-generated false statement more leniently than a human-generated one.
- › Omissions matter. AI is particularly likely to omit issuer-specific risks because it does not know the company's actual vulnerabilities.
- › Confidentiality and privilege are real risks. Feeding material non-public information into a public AI tool can waive privilege and create MNPI exposure.
Practical Risks in Offering Documents
AI-generated content creates specific risks in securities offerings:
- › Unsupported projections. A model can generate a five-year financial model that looks reasonable but is built on assumptions that do not match the company's contracts, pipeline, or historical performance.
- › Boilerplate risk factors. AI tends to produce generic risk language. The SEC routinely comments on risk factors that could apply to any company. Generic AI output makes that problem worse.
- › Inconsistent definitions. A model may use terms differently across sections of a document, creating internal contradictions.
- › Outdated law. Training data has a cutoff date. A model may cite rules, cases, or enforcement priorities that have changed.
- › Hallucinated citations. AI has been known to invent case names, statute citations, and SEC releases. Every citation must be verified.
A Defensible AI Policy for Issuers
Companies that use AI in the disclosure process should adopt a written policy that addresses:
- › Permitted uses. Which tasks may be assisted by AI and which require human drafting.
- › Verification requirements. Who reviews AI output and against what sources.
- › Confidentiality controls. What information may not be entered into AI tools.
- › Documentation. Maintaining records of what was AI-generated and what was verified.
- › Training. Ensuring that officers and staff understand that AI output is a starting point, not a finished product.
A policy alone will not prevent enforcement, but it demonstrates that the company takes its disclosure obligations seriously and does not treat AI as a substitute for legal counsel.
The Role of Securities Counsel
An experienced securities attorney does more than draft language. The attorney:
- › Identifies the facts that must be disclosed.
- › Evaluates whether a statement is materially misleading.
- › Coordinates with auditors, underwriters, and management.
- › Builds a record that can withstand SEC review and potential litigation.
AI can accelerate parts of that process, but it cannot perform the judgment at the center of it. The attorney's role is to ensure that the final document reflects the company's actual facts and risks, not a plausible approximation of them.
Practical Conclusion
AI is a tool, not a defense. The securities laws judge the document that is filed, not the tool that produced the first draft. Issuers that use AI responsibly — with verification, confidentiality controls, and lawyer oversight — can gain efficiency. Issuers that treat AI as a substitute for counsel are likely to discover that speed at the drafting stage becomes delay and exposure at the enforcement stage.
If your company is using AI to prepare an offering document or SEC filing, have a securities attorney review the output before it reaches investors or the Commission.
Related Reading
- › How the SEC Builds a Securities Fraud Case
- › Form S-1: Where the SEC Applies Scrutiny
- › The SEC Comment Letter Response Checklist
- › When Offering Optimism Becomes a Misleading Statement
<!-- related-links-injected -->
Authoritative Sources
- SEC Enforcement Manual, Division of Enforcement — U.S. Securities and Exchange Commission
- SEC Division of Enforcement — Litigation Releases — U.S. Securities and Exchange Commission
- 17 C.F.R. Part 240 — General Rules and Regulations, Exchange Act — Electronic Code of Federal Regulations
- SEC EDGAR — Full-Text Search of Company Filings — U.S. Securities and Exchange Commission
- Securities Act of 1933 (15 U.S.C. §§ 77a et seq.) — U.S. Government Publishing Office
- Securities Exchange Act of 1934 (15 U.S.C. §§ 78a et seq.) — U.S. Government Publishing Office
Primary sources are cited so readers can verify the law directly. Rules and staff guidance change; see our editorial and corrections policy to report an error or an outdated citation.
Frederick M. Lehrer served as an enforcement attorney in the SEC's Division of Enforcement at the Southeast Regional Office from 1991 through 2000, and concurrently as a Special Assistant United States Attorney in the Southern District of Florida from 1997 through 1999, prosecuting securities-related financial crimes. He has practiced securities and corporate law in private practice for more than twenty-five years, advising issuers worldwide on SEC registration, disclosure obligations, Regulation D private placements, Regulation A offerings, and going public transactions. The firm is based in Florida and serves clients internationally.