Liability for AI and software: updates to the European product liability regime

Last week, the European Parliament Committee on Internal Market and Consumer Protection (IMCO) published a draft opinion, which highlighted that AI applications could present challenges to the existing EU legal framework on product liability and called for further review by the European Parliament Committee on Legal Affairs (CLA).

In particular, the IMCO called on the CLA:

• to assess to what extent the Product Liability Directive needs to be updated to safeguard consumer welfare whilst avoiding unduly high costs for SMEs and start-ups;
• in particular, to consider whether software should be included in the definition of “products” and whether concepts such as “defect” in the existing framework should be updated to cater for characteristics of AI applications;
• to consider further harmonising liability rules in the EU to avoid fragmentation;
• to call on the EU Commission to evaluate whether and how the burden of proof should be reversed to empower harmed consumers, especially where third party software is involved or is part of a complex IoT environment; and
• ask the Commission to assess the need for a strict liability model for AI applications in high risk areas. 

Any possible changes to the product liability regime would be of interest to software developers and vendors, who would be brought within scope. The EU Medical Devices Regulation - implementation of which was delayed for a year due to the impact of Covid-19 - has already included certain medical software into the definition of “medical devices”, which are subject to more stringent regulatory scrutiny. 

The UK previously indicated that it would uphold its commitment to implement the MDR despite Brexit. Should the IMCO’s suggestions be adopted by CLA and the European Commission, the revised Product Liability Directive will become an important piece of regulation AI vendors and insurers will need to consider.