Product Liability and Safety_2026

USA – ILLINOIS Trends and Developments Contributed by: David A. Warnick, Johnson & Bell Ltd

Illinois Widens Reach of Product Liability for a New Generation of Products and Sellers The most significant product liability developments now taking shape in Illinois are two pending legisla - tive initiatives, each aimed at one of two threshold questions that have historically allowed defendants to limit or avoid exposure: what counts as a product, and who counts as a seller. One is directed at conver - sational AI, the other at online marketplaces. Together they reach manufacturers, distributors, and platform operators that have not traditionally regarded them - selves as classic product defendants. For companies that design, distribute, or sell goods into Illinois, these proposals matter even while they remain pending. They indicate where Illinois law is heading. They also build on an existing Illinois frame - work that heavily favours consumers. That preference has not gone unnoticed: a number of businesses have left the state in recent years, with the litigation cli - mate and the absence of tort reform among the fac - tors commonly cited. The pending bills would not so much create new exposure as extend existing liability to categories of defendant that have assumed they were outside its reach. For a defendant, these trends are best understood together rather than as separate developments. The product-versus-service distinction, the innocent-seller dismissal, and the challenge to personal jurisdiction have each given product defendants a pathway to dismissal. These emerging trends are likely to nar - row defendants’ ability to exit a case on procedural grounds. Conversational AI Enters the Strict Liability Framework Illinois has positioned itself among the first states that would treat conversational AI as a product subject to strict liability rather than as a mere service. Illinois House Bill 5044, the Chatbot Provider Liability Act, would deem chatbots to be products for the purpose of strict liability. It would impose on any chatbot pro - vider a duty to ensure that its system does not cause injury. The definition of provider is deliberately broad. It reaches anyone who creates, distributes, or makes a chatbot available. The duty attaches regardless of whether the provider exercised reasonable care, and

regardless of whether it had any direct contractual relationship with the injured user. The remedies the bill contemplates are as significant as the liability standard. A harmed user could recover actual damages and obtain injunctive relief. The user could also recover attorney’s fees and costs. Fee- shifting changes the economics of individual claims and makes a far wider range of disputes worth pur - suing. That is why the bill warrants attention even though it has since been re-referred to the House Rules Committee, the customary signal that a meas - ure has stalled for the session. The momentum has shifted to the Senate. Two companion measures there, Illinois Senate Bills 3502 and 3590, each styled an Artificial Intelligence Product Liability Act, remained active well after the House bill stalled, advancing to a third-reading deadline in the spring 2026 session. Senate Bill 3590 approaches the subject through the familiar vocabulary of product defect. It would author - ise product liability actions against the developer of an AI system for defective design, for inadequate instructions or warnings, and for failure to conform to an express warranty, and it would treat a deployer as a developer where the deployer makes a material and substantial change to the product or intentionally misuses it, contrary to an express warranty, in a way that proximately causes harm. The bill presses further on the burden of proof: a deployer of a high-impact system that fails to conduct documented testing and risk management would face a rebuttable presump - tion that the product is defective. These bills target a premise the technology sector has relied upon for years: that software is a service rather than a product. On that view, an “as-is” disclaimer and the ordinary negligence standard govern any dis - pute over a software output. However, strict liability removes that premise entirely. Once a chatbot is clas - sified as a product, the provider’s care in designing and distributing it becomes irrelevant to liability. The absence of a contract with the user also ceases to be a defence. The classification is what matters, and these bills make that classification by statute. The reach of this approach extends well beyond the small number of widely known developers that train

357 CHAMBERS.COM

Powered by