The U.S. Securities and Exchange Commission is putting corporate artificial intelligence claims under the microscope. The agency's Division of Examinations publishes annual priorities that signal where its exam staff will focus when reviewing registered firms, and on October 23, 2026, it said its fiscal year 2027 examination focus will include how accurately registered firms describe their AI capabilities. The message to the market is straightforward: investors should understand what they are buying when a company says "AI."

What the SEC Is Actually Looking At

There is no new AI disclosure rule here. The SEC has not written a dedicated rulebook for artificial intelligence. Instead, it plans to lean on the materiality frameworks it already has under Regulation S-K, the set of rules that governs what public companies must tell investors in their filings. Materiality, in securities law, turns on whether a reasonable investor would consider a statement important in weighing an investment decision — a long-standing standard that lets the agency test AI claims without waiting for new rules.

The target has a name: "AI-washing." The term describes companies overstating, inflating, or misrepresenting what their AI actually does, much as "greenwashing" became shorthand for exaggerated environmental claims. In practice, AI-washing can take many forms, from implying that a product runs on sophisticated models when it does not, to overstating the role AI plays in revenue or operations.

The SEC's stated tools are targeted comment letters and enforcement actions. Comment letters are the agency's way of asking a company to explain or fix a disclosure, and they typically press firms to substantiate what they have told the market; both the letters and the companies' responses are made public on EDGAR, the SEC's online filing system. Enforcement actions are what happens when the explanation is not good enough.

The examinations will also cover a second front: whether firms are prepared for AI-enhanced cyber threats. The SEC is effectively asking two questions at once — are companies honest about their own AI, and are they ready for someone else's?

The Numbers Behind the Scrutiny

The scrutiny arrives against a backdrop of extraordinary capital flowing into AI, and one filing illustrates the scale. Anthropic's recent IPO prospectus, its S-1, disclosed approximately $417 billion in future AI compute infrastructure obligations. That figure is a commitment, not revenue. It represents money the company has agreed to spend on the computing power that trains and runs its models.

Corporate America has noticed. Approximately 72% to 76% of S&P 500 companies now list AI as a material risk factor in their SEC filings — the section of a filing where a company spells out what could impair its business. Back in 2023, that figure was just 12%. In roughly three years, AI went from a niche risk mention to something most of the index feels obligated to flag.

Courtroom data helps explain why lawyers are nervous. AI-related securities lawsuits reached 16 cases in 2025, and the pace in 2026 suggests that total could double. These suits generally allege that misstatements about a company's business misled investors, which places the accuracy of AI claims squarely at issue.

What This Means for Companies and Investors

For companies, the immediate cost is compliance. Firms may need to invest more in legal review and risk management to make sure their AI disclosures hold up under examination. Legal and compliance teams are likely to spend more time reviewing how AI capabilities are described in public disclosures before regulators ask the same question.

Heavy spenders face a different challenge. When a company carries AI obligations measured in hundreds of billions of dollars, investors need to understand the assumptions behind those commitments and the risks if returns fall short. Anthropic's S-1 shows what that disclosure looks like at the frontier, and other companies making large infrastructure bets will likely face pressure to be similarly explicit about their obligations.

There is also a signal worth reading in the risk factor data. When roughly three-quarters of the S&P 500 call AI a material risk, the technology is no longer a side bet for most large companies.

Several signposts will show how the effort develops: the volume and content of SEC comment letters on AI disclosures, any enforcement actions tied to AI-washing, and whether the 2026 lawsuit count actually reaches the doubling pace.

The cyber angle deserves attention as well. As examiners probe readiness for AI-enhanced threats, firms that treat AI only as a revenue story may find the SEC equally interested in it as a security story.

Source: CryptoBriefing