Hunton Retail Law Resource

Recent product liability cases against A.I. companies are applying traditional product liability theories to a new technology.  

In February 2026, the California Superior Court for San Francisco County entered an order coordinating twelve cases pending against defendant OpenAI. See In re: ChatGPT Prod. Liab. Cases, Cal.Super. Ct., JCCP No. 5431. Plaintiffs allege that OpenAI’s ChatGPT is unreasonably dangerous and caused psychological harm to plaintiffs or their family members by reinforcing delusional beliefs, endorsing suicidal ideation and providing information to decedents about how to harm themselves, and contributing to users’ psychological deterioration. Plaintiffs further allege that OpenAI rushed ChatGPT to the market without adequate safety testing about the impact of the chatbot’s “sycophantic design” and lack of safety features on individuals’ mental health and physical safety.  

This consolidated action joins a growing wave of cases applying traditional product liability theories, including claims for strict product liability, negligence, failure to warn, and violation of state consumer protection statutes, to large language model technology. See, e.g., Garcia v. Character Techs., Inc., 785 F.Supp. 3d 1157, 1179 (M.D. Fla. 2025), motion to certify appeal denied, No. 6:24-CV-1903-ACC-DCI,2025 WL 2581834 (M.D. Fla. July 15, 2025) (denying in part motion to dismiss claims including strict liability, negligence, intentional infliction of emotion distress, unjust enrichment, and violation of Florida’s Deceptive and Unfair Trade Practices Act based on allegations that plaintiff’s son died by suicide because of his interactions with the company’s Character A.I. chatbot app); Joshi v. Open AI Foundation, Case No. 4:26-cv-00222 (M.D.Fla.) (bringing claims against OpenAI on behalf of a victim of an April 2025 shooting at Florida State University).   

The consolidated action against OpenAI in California, along with other pending product liability actions against A.I. companies across the United States, will need to address a number of novel issues specific to large language models. For instance, in Garcia, Character A.I. argued that its chatbot generated speech that its users have a right to receive under the First Amendment. While the court held that Character A.I. could assert the First Amendment rights of its users, it was not prepared to hold that Character A.I.’s output is speech that warrants First Amendment protections. See Garcia, 785 F. Supp. 3d at 1176.  

A.I. companies may also argue they are entitled to protection from product liability claims under Section 230 of the Communications Decency Act, which bars civil claims against interactive computer services based on the publication of third-party content. While social media companies have successfully relied on Section 230 as a defense to claims related to content posted on their platforms, plaintiffs are increasingly focused on specific design features that they argue falls outside the protections of Section 230.  

In Garcia, Character A.I. also argued that its chatbot is a service, not a product. The court determined that the chatbot was a product for the purposes of plaintiff’s product liability claims because the allegations focused on specific design features. Id. at 1180. The product v. service distinction has been an important test in recent social media product liability actions. While some courts have dismissed product liability claims against social media companies on the grounds that they provide a service rather than a product, others have allowed these claims to move past the pleading stage on the grounds that certain features or functionalities are more akin products. Compare Jacobs v. Meta Platforms, Inc., No. 22CV005233, 2023 WL 2655586, at *4 (Cal.Super. Mar. 10, 2023) (dismissing negligent design claim where court found “that, as a social media platform that connects its users, Facebook is more akin to a service than a product.”) with In re Soc. Media Adolescent Addiction/Pers. Inj. Prods. Liab. Litig., 702 F. Supp. 3d 809, 849 (N.D. Cal. 2023) (declining to treat platforms as a whole as products but evaluating “whether the various functionalities of defendants’ platforms challenged by plaintiffs are products”). Notably, while the consolidated proceedings against OpenAI are titled the “ChatGPT Product Liability Cases,” OpenAI’s filings in the consolidated proceedings characterize ChatGPT as a “a software-based service,” rather than a product.  

The A.I. product liability cases will likely open a new world of design defect arguments and analysis. Plaintiffs have argued that additional safeguards are necessary to protect users and flag concerning behavior. Notably, commentary from generative A.I. developers suggest that A.I. products might not fit the standard design defect analysis or feasible alternative design arguments because specific LLM output is not directly linked to any specific programing. As Dario Amodei of Anthropic explained in April 2025, “[m]odern generative AI systems are opaque in a way that fundamentally differs from traditional software. If an ordinary software program does something—for example, a character in a video game says a line of dialogue, or my food delivery app allows me to tip my driver—it does those things because a human specifically programmed them in. Generative AI is not like that at all. When a generative AI system does something, like summarize a financial document, we have no idea, at a specific or precise level, why it makes the choices it does—why it chooses certain words over others, or why it occasionally makes a mistake despite usually being accurate.” See Dario Amodei, The Urgency of Interpretability, April 2025. Discovery and expert analysis in these cases may reveal what additional design features and safeguards are feasible. 

We will be tracking the proceedings in ChatGPT Product Liability Cases closely.