Product liability and AI in Italy: risk profiles according to recent U.S. experience and EU prospects

published on 18 March 2025 | reading time approx. 4 Minutes

The evolution of legislation and case law on product liability raises crucial issues for companies developing and marketing AI-based software. The extension of product liability to software, recently introduced by Directive (EU) 2024/2853, and emerging trends in the United States highlight risks related to design defects, failure to provide adequate warnings, and damages resulting from the use of AI.

Directive (EU) 2024/2853: the extension of product liability to software

On November 18, 2024, the new Directive (EU) 2024/2853 was published in the Official Journal of the European Union, extending the traditional product liability regime — among other things — to software, addressing the challenges posed by the increasing spread of digital technologies and artificial intelligence (AI).

Pending the entry into force of the new European provisions, emerging litigation trends in the United States can provide significant insights into future cases related to liability for digital products and AI, as they already highlight potential risks for economic operators placing such products on the market.

In this regard, particular attention is drawn to the case Garcia v. Character Techs., Inc., No. 6:24-CV-01903, currently pending before the federal court of Florida.

The case

In October 2024, Ms. Garcia initiated proceedings before the federal court of Florida against various defendants, including the developer company of a well-known AI chatbot.

The claim seeks compensation for damages allegedly suffered due to the death of Ms. Garcia’s son, purportedly resulting from an addiction to the chatbot that led to his suicide.

The plaintiff has brought claims against various parties, including the chatbot developers, arguing that they designed and marketed a defective product lacking adequate safeguards for minors, further alleging the absence of appropriate warnings.

Applicability of the “Product Liability” regime

A preliminary and central issue in the present dispute concerns the legal classification of software as a "product" for the purposes of applying product liability rules. Previously, the chatbot in question would have been treated as a service.

By applying the defective product liability regime, however, a design defect has been alleged. Specifically, it is contended that the model underlying the chatbot was trained on an inadequate dataset, resulting in the generation of harmful and dangerous outputs for users, particularly minors.

Additionally, the lack of necessary warnings has been challenged, as the developers allegedly failed to provide adequate user guidance, allowing minors to interact freely with the software without proper safeguards.

The plaintiff has therefore brought legal action seeking compensation for damages arising from the minor’s death, including compensation for emotional distress suffered by herself.

Risk profiles for economic operators

The claim at issue—still pending but preliminarily deemed admissible by the federal court of Florida—although taking place within U.S. jurisdiction, offers valuable insights into the potential litigation risks faced by companies marketing AI-based products, including aspects related to compensation for loss of a relative and existential damages, as recognized under Italian law.

The above-mentioned risks underscore the importance for businesses operating in the AI sector to immediately implement stringent compliance measures and provide comprehensive and accurate information to mitigate litigation risks arising from the products they market—including software—and to ensure compliance with applicable regulations, particularly in view of the imminent applicability of Directive (EU) 2024/2853.

The evolving regulatory framework on product liability represents a fundamental shift in the legal treatment of software and AI-based technologies, with direct implications for companies operating in these sectors.

The U.S. experience, in a legal landscape already evolving for years, illustrates how the boundaries of liability for damages resulting from the use of these products are being defined.

Companies in this sector, as previously mentioned, will need to revise their compliance and security policies, implementing appropriate measures from an informational standpoint to prevent legal risks, avoid potential disputes, and maximize the opportunities offered by the expanding digital market.

Authors:

Giuseppe F. Bonacci

Stephan Daniel Cascone

Federico Pavesi