U.S. FEDResearch
SSRN: Geistfeld offers test for when AI outputs count as products under strict liability
In an October 12, 2026 SSRN preprint, Mark Geistfeld argues that courts deciding whether AI systems are products subject to strict liability or services governed by negligence need a theory of information. He distinguishes judgment-based content from representational content that is true or false, contending an AI output works like a product when its safety rests on an implied representation assessable without examining the provider's conduct.
Key points
- Geistfeld posted the article on SSRN on October 12, 2026.SSRN
- He says courts classify AI outputs by analogy to tangible products, an approach that lacks a theoretical foundation.SSRN
- He reads tangibility as a proxy for judging safety without assessing provider conduct and for distribution through a stream of commerce.SSRN
- Commercial distribution of AI systems ordinarily satisfies the stream-of-commerce criterion, he argues.SSRN
- A provider of representational content impliedly represents that it is true, much as a product seller represents reasonable safety.SSRN
- Without these accounts, he says, an AI companion allegedly fostering a teen's fatal dependency is indistinguishable from an AI story writer.SSRN