Tuesday, 02 June 2026 / Published in Artificial Intelligence, Law

AI Tort Liability: How Courts May Regulate Artificial Intelligence First *

Artificial intelligence regulation is often discussed as though it will emerge primarily through legislation or federal agency action.^[1] Congress debates AI bills. State legislatures consider safety mandates. International organizations publish ethical principles and governance frameworks. Yet another force is rapidly shaping the future of AI regulation, one that may ultimately prove more immediate and influential than formal legislation: tort litigation.^[2]

A growing body of legal scholarship and emerging case law suggests that courts are beginning to apply traditional product liability principles to artificial intelligence systems, particularly generative AI and chatbot platforms.^[3] The result may be a regulatory framework driven not first by lawmakers, but by judges, juries, litigants, and evolving tort doctrines.

Why AI Tort Liability Is Emerging Faster Than Legislation

Historically, tort law has repeatedly adapted to transformational technologies that introduced new forms of widespread risk.^[4] Automobiles, pharmaceuticals, medical devices, and mass-produced consumer products all forced courts to reconsider how liability should operate when technology causes foreseeable harm at scale.^[5] AI may now represent the next chapter in that evolution.

One of the central legal battles emerging in AI litigation involves a deceptively simple question: Is AI a product or a service? Traditionally, software companies have argued that digital systems primarily involve speech, information, or services rather than tangible products subject to strict product liability principles.^[6] Plaintiffs, however, are increasingly reframing AI systems as mass-marketed products whose design choices, safety features, warnings, and foreseeable risks should be evaluated under traditional product liability doctrines.^[7]

How Product Liability Principles Apply to AI Systems

That shift is significant because product liability law already contains mature legal frameworks for evaluating defective design, failure to warn, foreseeable misuse, safer alternative designs, and post market monitoring obligations.^[8] Courts do not need entirely new legal doctrines to begin addressing AI-related harms. Existing tort principles are proving adaptable to the technology.

Why Courts Are Focusing on AI Design and Safety Features

Recent litigation reflects this evolution. Courts increasingly appear willing to move beyond all-or-nothing arguments about whether AI platforms are categorically products or services. Instead, judges are examining discrete software features and asking whether those features function similarly to physical safety mechanisms.^[9] In litigation involving social media platforms, rideshare applications, chatbots, and digital platforms, courts have focused on issues such as parental controls, age gating, recommendation algorithms, friction-inducing prompts, escalation systems, and other safety-oriented design features.^[10]

This feature-by-feature approach may become one of the defining legal frameworks for AI litigation. Rather than litigating whether a chatbot’s output itself constitutes protected speech, plaintiffs are increasingly targeting the architecture of the system: the guardrails, defaults, warnings, monitoring systems, interaction patterns, and safety interlocks that shape user behavior.^[11] The legal focus is shifting from what did the AI say? to how was the system designed?

That distinction matters because courts are generally far more comfortable evaluating engineering choices and safety mechanisms than regulating abstract speech. Judges routinely oversee litigation involving defective designs, foreseeable risks, inadequate warnings, and safer alternative designs in countless other industries.^[12] AI litigation increasingly resembles those familiar forms of product litigation.

The Role of Tort Litigation in AI Governance

The implications for AI governance are substantial. Tort litigation naturally incentivizes companies to document testing procedures, conduct risk assessments, monitor post deployment behavior, evaluate foreseeable misuse, and implement safety-by-design principles.^[13] In product liability litigation, internal company records often become central evidence.^[14] Decisions about warnings, safeguards, testing timelines, user monitoring, escalation protocols, and alternative safety measures may all become discoverable evidence in future lawsuits.^[15]

This dynamic effectively creates a form of decentralized regulation through litigation pressure. Companies facing potential liability often alter practices long before legislatures finalize comprehensive regulatory schemes. Historically, product liability litigation influenced industries ranging from automotive manufacturing to pharmaceuticals, tobacco, and consumer electronics.^[16] AI companies may increasingly face similar incentives.

The Pharmaceutical Model and AI Risk Management

The analogy to pharmaceuticals is particularly instructive. Drug manufacturers operate within a framework that combines ex ante regulatory review with ex post tort liability.^[17] Even after approval, manufacturers continue monitoring products for emerging risks, updating warnings, and revising safety practices.^[18] Some legal scholars now argue that AI systems may require comparable forms of ongoing post market surveillance and risk management.^[19]

Importantly, tort litigation may also shape responsibility across the broader AI ecosystem. Emerging cases suggest plaintiffs will increasingly pursue claims not only against branded AI applications, but also against upstream foundation model developers, downstream deployers, enterprise integrators, and platform operators.^[20] Courts may ultimately apply cheapest cost avoider reasoning, imposing responsibility on the parties best positioned to reduce foreseeable harm.

At the same time, legislative and regulatory developments are beginning to reinforce these concepts. The European Union’s revised Product Liability Directive treats software and AI systems as products for liability purposes.^[21]California and other states are enacting AI-specific statutes focused on foreseeable risks, companion chatbots, and algorithmic safety.^[22] Even where such laws do not directly govern a case, they may influence how courts evaluate foreseeability, reasonableness, and standard-of-care arguments.

Why Courts May Become the Primary Regulators of Artificial Intelligence

The broader reality is becoming increasingly clear: AI governance may emerge through ordinary tort law faster than through comprehensive federal legislation. If that occurs, courts will occupy a central role in shaping the future boundaries of AI accountability. Judges will increasingly confront questions involving design defects, foreseeability, warnings, validation, human oversight, expert testimony, and algorithmic risk assessment.

Artificial intelligence is often described as a disruptive force for the legal system. Increasingly, however, the legal system itself may become one of the primary forces shaping artificial intelligence.

* This article was edited with the assistance of AI in the form of a large language model. It was used solely for grammar, editing, and footnote support. All substantive content and conclusions reflect human authorship.

^[1] Brian W. MacKenzie and David Wallace, AI in the Criminal Courts: Balancing Innovation and Justice, 41 Crim. Just.(Spring 2026).

^[2] Brian W. MacKenzie, AI Tort Liability: Does Negligence Law Still Apply?, Just. Speakers Inst. (2026).

^[3] Id.

^[4] Donald G. Gifford, Technological Triggers to Tort Revolutions, 95 N.C. L. Rev. 897 (2017).

^[5] Id.

^[6] Catherine M. Sharkey, Products Liability for Artificial Intelligence, Lawfare (Dec. 4, 2019).

^[7] Id.

^[8] Id.

^[9] Mark Geistfeld, Product Liability Law in the Age of AI (Aspen Publ’g 2025).

^[10] Id

^[11] Michael J. Lowell & Matthew J. O’Brien, AI Product Liability: The Next Wave of Litigation?, JDSupra (July 17, 2025).

^[12] Geistfeld, supra note 9.