What You Should Know
Numerous businesses today deploy artificial intelligence (AI) innovatively across various industries, with many companies eager to hop on the AI hype train and ride the momentum of new and innovative business tools. While regulatory oversight has been minimal, there are various industry guidelines, such as the NIST AI standards, on how companies should approach transparency in AI.
The Federal Trade Commission (FTC) is further laying the regulatory groundwork, through its recently launched Operation AI Comply, by taking enforcement actions against companies that make false, misleading, deceptive and unsubstantiated claims regarding their AI products and tools. Under Section 5(a) of the Federal Trade Commission Act, the FTC has broad enforcement power to prohibit unfair or deceptive acts or practices in or affecting commerce. The enforcement actions discussed below demonstrate that such deceptive practices are interpreted broadly, and that the FTC will hold companies accountable for their use of AI.
Recent FTC Enforcement Actions
DoNotPay, Inc.
In January 2025, the FTC settled an enforcement action against DoNotPay, Inc., an online subscription service for U.S.-based consumers claiming to offer an AI service that was “the world’s first robot lawyer.” The product was intended to help consumers with a range of commercial and legal matters, such as breach-of-contract issues, demand letters or release of liability waivers. The company claimed their product was trained and had expertise in over 200 areas of law. In fact, the FTC found, DoNotPay’s AI-powered product was not sufficiently trained on the body of federal and state laws or on how such regulations would apply to customers’ unique factual situations. DoNotPay also promised their AI chatbot was able to ask relevant questions regarding consumers’ legal needs and give consumers what they want, such as an “iron-clad demand letter.”
Contrary to these promises, the FTC found that the company’s chatbot was not trained or tested properly, as it did not ask relevant questions and its output resulted in ineffective documents. In addition, DoNotPay was also found to have engaged in unfair and deceptive practices against small business owners by claiming they could assess a business website for compliance and legal liability with just a business email address. The FTC found the representations of DoNotPay’s AI and other technology were misleading, unsubstantiated and exaggerated, and imposed a fine of $193,000 in addition to imposing ongoing advertising restrictions.
Evolv Technologies Holdings, Inc.
Evolv Technologies Holdings, Inc., a security screening company, marketed its subscription-based software and security equipment as AI-powered. The company claimed to have the ability to use proven artificial intelligence to detect threats. Customers were provided with metal detectors and could elect various sensitivity settings that would allegedly make it easier to discover potential weapons. But Evolv’s customers had a different experience – schools reported, for instance, that Evolv scanners failed to detect weapons and that alarms would be triggered by harmless objects. Even with higher sensitivity settings, false alarm rates increased. In November 2024, the FTC alleged that Evolv’s statements on how its AI and sensor technology could distinguish between personal items and weapons were inaccurate and false. In the settlement order that was approved in December 2024, Evolv will be banned from making unsubstantiated claims about its products’ detectability capabilities, and Evolv must also give certain of its K-12 school customers the option to cancel their contracts.
Rytr LLC
Rytr LLC is an AI writing assistant tool that generates business content, from reviews and testimonials to social media ads, blog posts and email responses. For business review generation, their users and subscribers could select the desired message tone, input keywords and level of creativity, and Rytr would generate content. However, the outputs were often highly specific and did not relate to the user’s inputs, resulting in deceptive and misleading reviews. If users posted the generated reviews online, they could have the impact of potentially deceiving customers into purchasing products/services mentioned in the reviews. The FTC filed a complaint against Rytr in September 2024, alleging that their generative AI writing assistant created fake content that enabled its users to engage in deceptive practices, and in December 2024 approved a final order against Rytr that prohibits the company from engaging in similar conduct and from advertising or selling any service generating reviews and testimonials.
IntelliVision Technologies Corp.
Another notable FTC case is against IntelliVision Technologies Corp., a facial recognition software using deep learning for original equipment manufacturers and consumers. IntelliVision claims their facial recognition software can identify faces of all ethnicities without racial or gender bias in their AI model. The company’s website states their AI model was trained on millions of images from across the world with a high accuracy rate. However, in its allegations the FTC stated that IntelliVision’s advertising and promotions of its facial recognition software were false and unsubstantiated as its software was trained on only approximately 100,000 faces. Moreover, IntelliVision’s product algorithms were not adequately tested, and they performed differently across demographics and genders. In January 2025, the FTC finalized an order settling these allegations where IntelliVision is barred from making false or misleading claims about its technology, and must rely on competent and reliable testing of its technology and submit to compliance reporting.
What You Can Do
Here are some actions that businesses using AI technology can take based on these recent FTC enforcement actions:
- It is just as important to be upfront about the limitations of your technology as it is to market the AI’s potential. Companies should prioritize transparency about the capabilities of their AI technology, training data and the various functionalities. Indeed, various AI laws coming into effect (such as California’s AB-2013) will require companies to be transparent in disclosing their training datasets and to log training data.
- Companies should take a closer look at their website policies. Businesses can bolster their credibility by scrutinizing their AI products’ privacy policies and terms of use to ensure that appropriate disclosures and disclaimers are made regarding their AI products.
- For any third-party AI products that a company uses, the company should ensure it has a complete understanding of the AI provider’s terms and how the AI provider uses the company’s data for training or improvement of the AI.
- Companies must have documented and robust AI governance policies that incorporate rigorous quality assurance testing of their AI tools and validation that they operate as promised.
- Organizations should safeguard against impermissible uses of their AI tools by educating potential customers on the AI technology and enforcing use restrictions in their terms of use and commercial contracts.
- Business entities should emphasize design principles resulting in fair, efficient and ethical AI throughout the development and deployment stages. Otherwise, companies can expose themselves to FTC scrutiny if they inadvertently become instrumentalities for their users to engage in deceptive practices.
- Companies should endeavor to work with independent third-party auditors that can conduct audits and testing to verify that AI tools and products perform as intended, are reliable and accurate, and comply with industry standards.
[View source.]
