Strict product liability traditionally applies to defective tangible goods and generally imposes liability without proof of negligence once a design, manufacturing, or warning defect is shown. Whether standalone software, an AI model, or a cloud-delivered service counts as a 'product' for these purposes, as opposed to a 'service' governed by ordinary negligence and contract principles, is a threshold question that can decide which legal theories are even available to a plaintiff — and courts across different jurisdictions have not landed on the same answer.

The classification question predates AI but has taken on new urgency as more consequential decisions run through software and models delivered as a service rather than shrink-wrapped code. Arguments for treating software like a product point to its mass-market distribution and the policy rationale behind strict liability — spreading the cost of defects to the party best positioned to prevent them; arguments against point to software's intangibility, its frequent delivery bundled with ongoing services and updates, and courts' traditional reluctance to extend strict liability doctrine built for physical goods.

Juricratic models this classification as a threshold dial that shapes which liability theories are even in play for a given matter, rather than resolving it — since the honest position is that the answer differs by jurisdiction and by exactly how the software or model was delivered, and that variance is itself decision-relevant information for counsel.

Counsel evaluating a software-harm claim researches how the relevant jurisdiction has historically classified analogous software or SaaS delivery models before committing to a strict-liability theory, since pleading the wrong classification can result in early dismissal of that count. Where the law is unsettled or split, plaintiffs often plead strict liability and negligence in the alternative so the case can proceed regardless of how the classification question is ultimately resolved.