Liability

AI Washing: How Boards Can Spot It and Why It Matters for Director Liability

Paula Fontana April 2026 7 min read

When a company overstates what its AI does, directors can find themselves personally exposed, especially if they approved communications they didn't scrutinize. AI washing is moving from a reputational risk to a legal one, and the standard of director oversight is moving with it.

The term "AI washing" borrows its structure from greenwashing, the practice of making claims about environmental commitment that outpace the underlying reality. In the AI context, it refers to companies overstating what their AI systems actually do. Claiming autonomous capabilities that are largely manual. Describing machine learning where there is no learning. Asserting accuracy rates that have not been independently validated.

For a long time, AI washing was primarily a reputational risk. Embarrassing if caught, damaging to credibility, but not necessarily a legal matter. That is changing rapidly.

Regulators Are Paying Attention

The SEC has been explicit about AI washing as a securities fraud concern. When a company makes materially misleading statements about its AI capabilities to investors, those statements fall within the SEC's disclosure enforcement mandate. Several enforcement actions have already been brought. More are expected.

State consumer protection regulators are developing parallel frameworks. Healthcare regulators are scrutinizing AI capability claims in clinical and administrative contexts. The pattern is consistent. The gap between what companies claim their AI does and what it actually does is becoming a legal exposure point.

Why Directors Are Exposed

Board directors typically did not write the press release. They did not design the investor presentation. They were not in the room when the marketing language was drafted. They approved the annual report. They signed off on the proxy statement. They received and reviewed the earnings script before the call.

Under current fiduciary duty standards, directors are expected to exercise reasonable oversight over material communications, including communications about AI capabilities. The defense that the board relied on management's representations is weaker in the AI context than it has historically been in others. The expectation that boards ask substantive questions about AI claims is now documented in regulatory guidance, governance frameworks, and published standards.

The director who asks whether AI claims have been independently validated and receives a documented answer is in a better position than the director who approved the communication without asking.

The Questions to Ask Before Approving AI-Related Communications

Boards do not need to become AI experts to discharge this oversight responsibility. They need to ask the right questions and ensure that management's answers are documented.

  • What specifically does the AI system do? Phrases like "AI-powered" and "machine learning-driven" describe a category of system without specifying its actual function. Get to the actual function.
  • Have these capability claims been independently validated? Who validated them? What methodology did they use? When was the validation last updated?
  • Are the accuracy or performance metrics being cited current? AI systems degrade over time. A performance metric from deployment may not reflect current performance.
  • Does the claim accurately represent the role of human oversight? Many systems described as "automated" involve significant human review. Claims that omit that context can mislead.
  • Has legal or compliance reviewed the AI-related language specifically? General legal review of a communication does not necessarily include expert review of the AI claims within it.

The Reputational Dimension

Beyond the legal exposure, AI washing creates a governance credibility problem that is increasingly hard to recover from. Institutional investors, proxy advisors, and governance rating agencies are developing AI-specific oversight criteria. A board that approved overstated AI claims faces questions about the quality of its oversight that will surface in future proxy seasons, governance reviews, and investor conversations.

The safeguard is straightforward. Build the habit of scrutiny around AI claims the same way mature boards have built it around financial projections. The underlying discipline is identical. Only the subject matter is new.

Ready to assess your board?

The Compass is where every Vela engagement begins. Request access or a briefing to see what it surfaces for your board.

Request a Briefing More Resources