Loading
AI washing is the practice of overstating, fabricating, or obscuring the true extent of artificial intelligence in a product, service, or company's operations when marketing to investors, customers, or regulators. The term moved from industry jargon to formal enforcement language in March 2024, when the US Securities and Exchange Commission brought its first-ever cases against investment advisers for AI-related misstatements, and the pattern has since expanded to public companies, startup founders, and consumer products. It matters for governance because AI washing is no longer just a reputational risk, it is an active enforcement theory under securities-fraud, marketing-rule, and consumer-protection law in the United States, with Australian and other regulators signalling similar concern. As agentic AI hype accelerates, a closely related pattern called "agent washing", rebranding chatbots, robotic process automation, and simple workflow tools as autonomous agents, has emerged as the next front in the same regulatory logic.
Run the free AI Health CheckAI Washing, the practice of misrepresenting the AI capability, sophistication, or involvement of a product, service, or company to investors, customers, or regulators.
AI washing has become an enforcement priority. The SEC has charged multiple companies for materially misleading AI claims, including Delphia (USA) Inc. and Global Predictions Inc. in March 2024 (the first-ever AI-washing actions, with $225,000 and $175,000 penalties), Presto Automation Inc. in January 2025 (the first action against a public company), and Albert Saniger of Nate Inc. in April 2025 (parallel SEC civil and DOJ/SDNY criminal charges). The FTC has acted under Section 5, notably through its February 2023 guidance "Keep your AI claims in check" and its September 2024 enforcement sweep Operation AI Comply, which covered DoNotPay, Rytr, and several AI-branded business-opportunity schemes. The ACCC has flagged AI-washing as an emerging Australian Consumer Law risk rather than a declared enforcement priority, and Gartner has coined "agent washing" specifically for the practice of rebranding chatbots and RPA as agentic AI, estimating only ~130 of thousands of "agentic AI" vendors offer genuine agentic capability.
Source: SEC v. Delphia/Global Predictions (2024), Presto Automation (2025), Saniger/Nate Inc. (2025); FTC "Keep your AI claims in check" (2023) and Operation AI Comply (2024); ACCC AI transparency statement (Feb 2026)
In the United States, AI washing sits at the intersection of two enforcement regimes. For investment advisers and public companies, the SEC has used the antifraud provisions of the Investment Advisers Act (Sections 206(2) and 206(4)), the Marketing Rule (206(4)-1), and the Compliance Rule (206(4)-7) to charge firms whose AI claims in filings, ads, or investor communications did not match what their systems actually did. For public companies, the theory extends to disclosure obligations under the Securities Act and Exchange Act, a misleading claim about a flagship AI product's capability can be treated the same as any other materially false statement to investors.
The Federal Trade Commission relies on Section 5 of the FTC Act, which prohibits unfair or deceptive acts or practices. Its February 2023 guidance, 'Keep your AI claims in check,' put companies on notice that merely using an AI tool somewhere in development is not the same as a product 'having AI in it,' that performance claims need proof, and that claims a system can do something beyond current AI capability are treated as deceptive. In September 2024 the FTC operationalized that guidance with 'Operation AI Comply,' a coordinated sweep of enforcement actions against companies making unsubstantiated AI claims.
Delphia (USA) Inc. & Global Predictions Inc., March 2024
The SEC's first-ever AI-washing cases. Delphia claimed it used AI and machine learning on clients' spending and social-media data to inform investment advice when it did not; Global Predictions claimed to be the 'first regulated AI financial advisor' offering 'AI-driven forecasts' it could not support. Both settled, paying $225,000 and $175,000 in civil penalties respectively, about $400,000 combined, plus Marketing Rule and Compliance Rule violations.
Presto Automation Inc., January 2025
The SEC's first AI-washing action against a public company. Presto overstated its drive-thru voice-ordering AI, failing to disclose that most of the underlying speech technology was built and run by a third party, and downplaying how often human staff had to intervene (in SEC filings, non-intervention rates were shown well below what was implied). Presto consented to a cease-and-desist order with no penalty, citing its financial condition.
Albert Saniger / Nate Inc., April 2025
Parallel SEC civil and DOJ/SDNY criminal charges against the shopping-app founder, who raised over $42 million by claiming Nate's app used AI to autonomously complete online purchases. Regulators allege the orders were actually processed manually by contract workers overseas. This was the first AI-washing case brought under the current administration, and the first to pair SEC charges with a parallel criminal indictment.
FTC 'Operation AI Comply', September 2024
A coordinated sweep targeting AI-related deception, including DoNotPay (marketed as the 'world's first robot lawyer,' later ordered to pay $193,000 and notify affected subscribers), Rytr (an AI writing tool the FTC said let customers mass-generate fake reviews), and several 'AI-powered' business-opportunity schemes accused of false earnings claims.
As agentic AI became the dominant vendor narrative, Gartner identified a related practice it calls 'agent washing', vendors rebranding existing chatbots, robotic process automation, and rules-based workflow tools as 'agentic AI' without the underlying autonomy. In a June 2025 press release, Gartner estimated that only about 130 of the thousands of vendors marketing agentic AI products offer genuine agentic capability, based on a poll of over 3,400 webinar attendees. The same release projected that more than 40% of agentic AI projects will be cancelled by the end of 2027 due to escalating costs, unclear ROI, or inadequate risk controls.
Agent washing raises the same governance questions as AI washing generally: does the product do what the vendor says, is 'autonomous' or 'agentic' a substantiated technical claim or a marketing label, and would a regulator's review of the system's actual decision-making match the pitch deck?
Substantiate before you publish
Every AI capability claim in a filing, pitch deck, ad, or website should be traceable to evidence, model documentation, test results, or architecture review, before it goes out, not after a regulator asks.
Match filings to production reality
If a system relies on a third-party model, manual human review, or a fallback process, disclose it. The Presto case turned on undisclosed reliance on a third party and understated human intervention rates.
Treat 'agentic' and 'autonomous' as specific claims
Gartner's agent-washing findings suggest regulators and buyers alike will start testing whether a system meaningfully plans and acts without human steps, not just whether it uses an LLM.
Align legal, product, and marketing sign-off
The SEC's Compliance Rule charges in the Delphia and Global Predictions cases specifically penalized the absence of internal policies to catch AI-related misstatements before they reached investors.
What is AI washing?
AI washing is misrepresenting how much artificial intelligence a product, service, or company actually uses or how capable that AI is, when communicating with investors, customers, or regulators. It covers everything from claiming AI-driven processes that are actually manual, to exaggerating a real AI system's accuracy or autonomy.
Is AI washing illegal?
It can be. In the US, the SEC has charged AI washing as securities fraud and Investment Advisers Act violations (antifraud, Marketing Rule, Compliance Rule provisions), and the FTC has charged it as an unfair or deceptive practice under Section 5 of the FTC Act. Penalties have included civil monetary fines, cease-and-desist orders, and in one case a parallel criminal indictment.
What was the first SEC enforcement action for AI washing?
In March 2024 the SEC settled its first-ever AI-washing cases against investment advisers Delphia (USA) Inc. and Global Predictions Inc., which paid $225,000 and $175,000 in civil penalties respectively for falsely claiming AI-driven investment processes they did not actually have.
Has the SEC charged a public company with AI washing?
Yes. In January 2025 the SEC charged Presto Automation Inc., a formerly Nasdaq-listed restaurant-technology company, for overstating the capabilities of its AI drive-thru ordering product and failing to disclose reliance on a third party and high rates of human intervention.
What is 'agent washing' and how is it different from AI washing?
Agent washing is a specific form of AI washing aimed at the current agentic AI trend: vendors relabeling ordinary chatbots, robotic process automation, or scripted workflows as autonomous 'AI agents.' Gartner estimated in June 2025 that only around 130 of the thousands of vendors marketing agentic AI products offer genuine agentic capability.
How can a company avoid AI-washing risk?
Substantiate every AI capability claim with evidence before publishing it, make sure filings and marketing match what the system actually does in production (including any human-in-the-loop steps or third-party components), and put cross-functional sign-off between legal, product, and marketing on AI-related claims, the absence of exactly this kind of internal control was itself charged as a violation in the SEC's first AI-washing cases.
Last reviewed July 2026