When AI gets it wrong: Insurance coverage for false, defamatory and infringing AI-generated content

Generative artificial intelligence cannot be avoided or ignored, but its risks can be managed. As organizations increasingly deploy Gen AI in marketing, customer interactions and operational decision-making, they are also assuming risk for the outputs those systems generate.

AI systems have already produced statements and content that are false, misleading, defamatory or infringing, exposing companies to a growing range of legal claims. In some cases, AI tools have generated incorrect statements about identifiable individuals. In others, AI systems have produced text, images or other materials that allegedly incorporate copyrighted works or personal data. While such content may result in claims alleging familiar legal theories, the resulting claims raise new questions about who is responsible when the harmful content is generated by an AI system rather than a human actor.

These claims raise an increasingly important question for risk managers and policyholders: which insurance policies respond when liability arises from AI-generated outputs? Existing insurance programs may provide coverage in some circumstances, but insurers are increasingly introducing exclusions and limitations aimed specifically at AI-related risks.

Claims arising from false, misleading, defamatory or infringing AI-generated outputs

Liabilities arising from the use of artificial intelligence are no longer hypothetical. Increasingly, claims arise when organizations rely on AI-generated outputs that later prove to be inaccurate, unlawful or harmful. Those claims may involve false statements about individuals, AI-generated content that incorporates copyrighted works, or outputs that rely on personal data. The legal theories themselves are familiar, but the alleged misconduct originates from AI-generated outputs that may be false, misleading, defamatory or infringing.

1. Defamation and reputational injury. One emerging area of litigation involves defamation and reputational injury caused by AI-generated statements. For example, in 2025, radio host Mark Walters sued OpenAI after ChatGPT falsely stated that Walters had been accused of embezzlement in a legal complaint. The court dismissed the case, concluding that Walters failed to show actual malice or negligence. While that decision favored the AI developer, it illustrates the type of claim that may arise when AI systems generate false statements about identifiable individuals. A related and rapidly emerging category of risk involves so-called "deepfakes," in which artificial intelligence systems generate realistic images, audio or video depicting individuals saying or doing things that never occurred. Deepfakes have already been used to fabricate political statements, impersonate corporate executives and create misleading videos of public figures. When such content falsely portrays identifiable individuals, it may give rise to claims for defamation, false light or misappropriation of name or likeness. As generative AI tools make it easier to produce convincing fabricated media, organizations that publish, distribute or rely on such content may face increasing litigation risk.

2. Intellectual property and AI-generated content. Another rapidly expanding area of litigation involves copyright and intellectual property claims related to AI-generated content. Several publications and companies are suing popular Gen AI developers for alleged use of their copyrighted material to train their systems. These disputes often center on whether AI systems generate outputs that reproduce or closely resemble copyrighted works, creating potential infringement liability for those who publish or distribute the resulting content. While these suits primarily focus on AI developers, even those not in the business of developing AI are at risk of liability if they inadvertently publish or distribute AI-generated content that incorporates copyrighted works. As organizations increasingly rely on AI to generate marketing materials, website content and advertising copy, disputes may arise over whether existing insurance policies respond to claims alleging copyright infringement based on AI-generated content.

3. Privacy and data-use claims. Data privacy claims represent another significant front of litigation related to Gen AI. As one recent example, a proposed class attempted to sue OpenAI and Microsoft for using stolen personal data to train ChatGPT and other Gen AI tools. Even companies that do not develop AI models may face liability where AI systems process or generate outputs based on personal data. In those situations, liability may arise when AI-generated outputs reveal or rely upon personal data in ways that violate privacy laws or data-use restrictions. These claims may arise under a wide range of privacy statutes, including state consumer privacy laws, biometric privacy statutes and federal statutes governing data collection and consumer reporting. As AI practices come under greater scrutiny, data privacy claims are likely to become one of the most active fronts of litigation in the Gen AI space.

4. Discrimination. The use of AI systems to make employment and business decisions has also created similar litigation risks based on the same discriminatory concepts observed in human decision-making. In 2023, in the first case of its kind, the Equal Employment Opportunity Commission (EEOC) settled a case with iTutorGroup, Inc. over alleged discriminatory hiring practices. The EEOC alleged that iTutorGroup's AI-hiring software automatically rejected older job applicants in violation of the Age Discrimination in Employment Act.  More recently, lawsuits have been filed challenging AI-based hiring and screening tools used by numerous employers. These claims often allege that AI systems replicate or amplify discriminatory patterns present in historical data. In many cases, the alleged discrimination arises from AI-generated recommendations or screening results that employers rely upon when making hiring or employment decisions. These cases highlight that use of AI can expose companies to the same types of discrimination liabilities long associated with human decision-making.

5. Bodily injury and property damage. AI-generated outputs may also create liability when false or misleading statements produced by AI systems lead to physical harm. Unlike traditional product liability claims involving defective machines, these cases arise when users rely on AI-generated advice, recommendations, or instructions that prove inaccurate or dangerous. For example, lawsuits have been filed alleging that AI chatbots generated statements encouraging harmful behavior or provided misleading guidance about dangerous activities. One widely reported case involves the family of a teenager who tragically died by suicide and alleged that an AI chatbot encouraged self-harm.  The lawsuit asserts claims for wrongful death, product liability and failure to warn. These cases illustrate that the risk does not arise solely from malfunctioning technology, but from AI-generated statements that users interpret as reliable information or guidance. As organizations increasingly deploy AI systems in customer-facing tools and advisory functions, disputes may arise when false or misleading AI-generated statements contribute to physical injury.

6. Securities fraud. Another emerging front for AI liability is securities claims. Increasingly, companies face exposure for how they describe their use of the technology. Known as "AI washing," making false or misleading statements about the integration of AI can draw actions from the SEC, the DOJ and investors or shareholders. In 2024, two investment advisers, Delphia Inc. and Global Predictions Inc., paid $400,000 in penalties to the SEC for allegedly making false and misleading statements about their use of AI.  In 2025, the SEC and DOJ filed actions against the former CEO of Nate Inc., a tech startup, alleging he made false statements to investors about the company's purported use of AI technology to raise over $42 million.

As technology continues to evolve, so too will the potential exposure.

Existing "silent" coverage and limitations

One might assume that broad liability coverage already in place would guard against claims arising from false, misleading, defamatory or infringing AI outputs. To some extent, that assumption is correct: Existing insurance policies likely cover some, if not most, of the potential liability companies could face from the implementation and utilization of Gen AI. However, existing policies likely are imperfect solutions and coverage for AI-related claims will often depend heavily on how the claim is framed. Insurers may argue that the alleged harm did not arise from the insured's own statements or conduct, but instead from autonomous outputs generated by an artificial intelligence system, creating new disputes over whether traditional liability coverage applies.

For example, one of the most common and expansive types of insurance is commercial general liability (CGL) insurance. CGL policies often have broad coverage for losses involving bodily injury, property damage, and personal and advertising injury. Personal and advertising injury coverage, in particular, may be implicated where AI-generated statements are alleged to be defamatory or where AI-generated marketing materials incorporate copyrighted content. However, CGL policies frequently exclude coverage for losses directly or indirectly resulting from assault and battery, which could implicate scenarios like the teen suicide and abuse cases or other incidences. CGL policies often also contain cyber and data exclusions that could exclude coverage for data privacy losses and breaches.

Directors and officers (D&O) policies are also extremely common, and will often cover violations of securities laws, which could include AI washing. However, coverage for the insured entity may be limited to securities claims, and D&O policies may exclude third-party bodily injury, property damage and employment claims.

Employment practices liability (EPL) policies, on the flip side, generally offer broad coverage for allegations of wrongful employment practices, including many types of discrimination. These policies, however, may limit or exclude coverage for certain privacy-related claims.

Cyber or technology errors and omissions insurance generally offer coverage for losses stemming from data security, breaches and attacks. These policies also typically will not exclude coverage of AI. However, they are also among the most variable lines of coverage, differing significantly from one policy form to another and from carrier to carrier, and in many cases, may provide only narrow protection for specific categories of third-party claims.

New and emerging limitations and exclusions

Existing policies may respond to some of the risks created by Gen AI, but they are far from complete solutions. Just as the technology and liability landscape relating to AI is rapidly evolving so too is the insurance market. On one side of the coin, carriers are introducing new exclusions and limitations to address Gen AI, with some excluding AI risks from policies altogether. On the other side, as these exclusions appear, insurers are introducing new solutions to provide additional coverage options.

Some notable limitations and exclusions come from Philadelphia Indemnity Insurance Co., which recently added exclusions to some of its policies to except advertisements and content created by Gen AI; Hamilton Select Insurance, which recently excluded claims based upon or arising out of actual or alleged use of Gen AI; and Berkley, which appears to be implementing absolute AI exclusions across D&O, E&O and fiduciary liability policies. More recently, the Insurance Services Office, which develops widely used standard-form policy language for insurers, introduced several variations of exclusionary endorsements for CGL policies targeting generative AI claims. In its broadest form, the exclusion will eliminate coverage altogether for claims arising out of generative AI where any causal connection to AI is alleged.

At the same time, other insurers like AXA, Coalition and Munich Re, to name a few, are rolling out policies that are very inclusive of AI, but may come at an additional cost. Startups like Armilla, backed by Lloyds of London, are offering product warranties on AI solutions by partnering with insurance providers.

Takeaways

As companies increasingly rely on AI-generated outputs in marketing, customer interactions and operational decision-making, the line between human-generated and machine-generated content will continue to blur. That dynamic will inevitably lead to new disputes over responsibility for AI-generated harm and over which insurance policies respond.

Organizations that proactively evaluate these risks and address potential coverage gaps today will be better positioned to capture the benefits of AI while managing its evolving liabilities.