13 July 2026
The last six months have confirmed one clear global trend: regulators everywhere are racing to bring software and AI inside the definition of a “product.” The EU is furthest along with a binding directive; the UK is consulting; the US is fighting the question out in the courts and statehouses; and India is focused more on e-commerce accountability than on AI-specific liability for now. Below is a side-by-side look at what changed, what’s coming, and what businesses should be watching.
No comprehensive federal “product liability modernization” law exists, so change is happening through courts, state legislatures, and agency enforcement rather than a single statute.
• Courts are redefining “product” function-first. In the closely watched social-media addiction MDL, a federal court rejected the argument that platforms categorically fall outside product liability, instead analyzing alleged defects feature-by-feature, a sign that software and AI-driven services are losing their historic immunity from strict liability.
• State AI laws vs. federal pushback. In 2025, 38 states enacted some form of AI regulation. Colorado’s AI Act (in effect February 2026) is the first to require companies deploying “high-risk” AI systems to assess harms and disclose automated decision-making. At the same time, a federal executive order created an AI Litigation Task Force to challenge state AI laws deemed unconstitutional or burdensome, setting up a federal-state collision.
• CPSC enforcement is intensifying despite leadership turnover. Recalls and civil penalties for late hazard reporting hit record levels in 2025 and continue into 2026; mandatory electronic filing under Section 15 takes effect July 8, 2026.
• Pricing and “dark patterns” are now product-liability-adjacent. The FTC, 21 states, and D.C. sued Uber over its subscription cancellation flow; several states have passed laws requiring disclosure when prices are set algorithmically, and California has introduced civil and criminal liability for unfair algorithmic pricing.
• Autonomous vehicles remain a liability blind spot. SAE Level 4 robotaxis are expanding into more cities with no uniform federal liability framework for crashes involving no human driver.
Watch next: Whether Congress attaches AI preemption language to a spending bill, and how state AGs respond.
The EU has gone furthest, fastest, with a binding, harmonizing directive that reaches nearly every business selling connected products into Europe.
• New Product Liability Directive (EU) 2024/2853 entered into force in December 2024 and must be transposed by every Member State by 9 December 2026. It applies only to products placed on the market after that date.
• Software and AI are now explicitly “products.” Standalone software, AI systems, and digital manufacturing files fall within strict, no-fault liability for the first time. Free open-source software remains excluded, but only if developed outside commercial activity.
• Claimant-friendly evidentiary shift. Courts may order disclosure of technical files, and defectiveness/causation can be presumed where a claimant faces “excessive difficulty” due to technical complexity, language written with AI’s “black box” problem specifically in mind.
• Damages caps removed. The old €500 threshold for property damage and financial liability caps for personal injury are gone, and recoverable harm now includes data loss/corruption and recognized psychological harm.
• AI Act becomes the liability yardstick. Non-compliance with AI Act safety obligations (accuracy, robustness, cybersecurity, data governance) will now directly inform whether a court finds an AI product “defective” under the new Directive.
• National transposition is already diverging. As of April 2026, Germany, the Netherlands, Denmark, Finland, Czechia, Slovakia, and Sweden have published transposition bills; Hungary has already adopted its law; France and Italy have not yet transposed. Germany’s draft, for instance, carves out medicinal products entirely, keeping them under separate strict-liability rules.
Watch next: Whether national transpositions stay harmonized or fragment the single market by December’s deadline.
Post-Brexit, the UK is deliberately choosing how much of the EU’s approach to mirror, rather than adopting it automatically.
• Product Regulation and Metrology Act 2025 (Royal Assent 21 July 2025) gives government broad powers to update product safety rules via secondary legislation, covering the full product lifecycle, but makes no substantive changes on its own.
• Law Commission’s first review since 1987. Launched days after the Act passed, this full-scale review of the Consumer Protection Act 1987 is examining whether the regime is fit for digital products and AI. A formal public consultation is expected in the second half of 2026, with recommendations to government likely late 2026/early 2027.
• Government consultation on replacing the General Product Safety Regulations 2005 opened 31 March 2026 and runs to 23 June 2026, described as the most significant overhaul of UK product safety in two decades. It proposes AI-enabled product regulation, digital-by-default labelling, and a call for evidence on balancing AI safety with innovation.
• Online marketplace accountability is coming, following the government’s 2024 commitment to hold marketplaces responsible for dangerous third-party products sold on their platforms, though secondary legislation timing remains unclear.
• AI-specific legislation remains stalled. A private member’s Artificial Intelligence (Regulation) Bill has been reintroduced but has not progressed; the UK’s approach to AI remains comparatively light-touch versus the EU.
Watch next: Whether the Law Commission’s eventual recommendations pull UK product liability law closer to, or further from, the EU’s new Directive.
India’s most active front right now is digital marketplace transparency and dark-pattern enforcement, built on the Consumer Protection Act, 2019 rather than new AI-specific liability rules.
• Legal Metrology (Packaged Commodities) Amendment Rules, 2026 (notified February and April 2026) impose new e-commerce disclosure duties — most notably mandatory Country of Origin labelling on online listings, closing a gap that exists in physical retail but not digital retail.
• Dark patterns enforcement continues. The Central Consumer Protection Authority (CCPA), which recognizes 13 specific dark patterns (false urgency, subscription traps, drip pricing, and others), has pushed major platforms — including large ticketing and travel players — to self-audit and self-certify they don’t use them.
• Product liability under Chapter VI of the CPA 2019 remains India’s core statutory basis for defect-based claims, with strict liability available where a manufacturing defect, design deviation, or inadequate warning is shown — but relief is only available to those who meet the Act’s definition of “consumer” (excluding commercial/resale buyers).
• Consumer commissions continue to expand accountability, including holding both manufacturers and dealers jointly liable in vehicle defect cases and treating one-sided contract terms that minimize manufacturer liability as “unfair contracts” under the Act.
• No AI-specific product liability framework yet — India’s product liability regime has not been amended to address AI/software the way the EU’s has, making this a notable point of divergence.
Watch next: Whether fallback liability proposals for e-commerce platforms (making platforms liable for seller defaults) are finally formalized, after being under discussion in various forms since 2021.
| Theme | US | EU | UK | India |
|---|---|---|---|---|
| Is software/AI a “product”? | Contested, decided on a case-by-case basis in court | Yes, explicitly (from Dec 2026) | Under review (Law Commission) | Not yet addressed |
| Central reform vehicle | State laws + litigation | Directive 2024/2853 | PRAM Act 2025 + consultation | CPA 2019 + Legal Metrology amendments |
| Key 2026 deadline | July 8 (CPSC e-filing) | Dec 9 (transposition) | June 23 (consultation closes) | Ongoing rulemaking |
| Main current focus | AI liability, dynamic pricing, CPSC enforcement | AI Act-linked defect standard, disclosure rules | Product safety framework overhaul | E-commerce transparency and dark patterns |
If you sell connected products, software, or AI-enabled goods into more than one of these markets, the EU’s December 2026 deadline is the forcing function, it is setting the de facto global benchmark that the UK is watching closely and that US litigants are already citing by analogy. Companies should be mapping which products will be “placed on the market” before versus after 9 December 2026, and treating technical documentation and AI Act compliance as liability evidence, not just regulatory paperwork.
