On January 23, 2020, the European Parliament’s Internal Market and Consumer Protection Committee approved a resolution on artificial intelligence (“AI”) and automated decision-making (“ADM”). The resolution references several major pieces of work carried out by the European Commission on AI and provides a list of existing EU instruments that are relevant to AI and ADM — which together present a potential roadmap of areas of reform.
The resolution was approved in committee by 39 votes in favor, none against and four abstentions. It will next be voted on by the full Parliament in an upcoming plenary session. If adopted, it will be transmitted to the EU Council and the Commission for consideration. The Commission’s Executive Vice-President Margrethe Vestager is expected to present plans for a European approach to AI in a Commission meeting on February 19, 2020.
Among other things, the resolution states that:
- When consumers interact with virtual assistants and chatbots, they should be informed about how the system functions, how to reach a human with decision-making powers, and how the system’s decisions can be checked and corrected;
- A risk-based approach to regulation should be taken, in light of the varied nature and complexity of the challenges created by different types of AI and ADM;
- ADM systems should use “high-quality and unbiased datasets” and “explainable and unbiased” algorithms; and
- Businesses should have internal review structures to remedy mistakes in automated decisions, and that consumers should be able to seek human review of, and redress for, automated decisions that are final and permanent.
The resolution also stresses the importance of examining the current EU legal framework — in particular, in the fields of consumer protection, data protection, and product safety and market surveillance. Among the specific EU measures highlighted are:
- Better Enforcement Directive (Directive (EU) 2019/2161) — the resolution urges the Commission to closely monitor the implementation of the Diretive’s new rules requiring traders to inform consumers when prices of goods or services have been personalized on the basis of ADM, and when the profiling of consumer behavior allows traders to assess the consumer’s purchasing power.
- Alternative and Online Dispute Resolution (Directive 2013/11/EU and Regulation (EU) No 524/2013) — the resolution calls on the Commission that any upcoming review of the Directive or Regulation on alternative and online dispute resolution for consumer disputes takes into account the use of ADM and ensures that humans remain in control.
- Specific Product Safety Rules (e.g., Machinery Directive, Toy Safety Directive, General Product Safety Directive, etc.) — the resolution calls on the Commission to adapt product safety rules so as to ensure that they provide clarity to manufacturers about their obligations, and that users have clarity on how to use products with ADM capabilities.
- Product Liability Directive (Council Directive 85/374/EEC) — the resolution urges the Commission to review the Product Liability Directive and consider updating concepts such as “product”, “damage” and “defect” and the burden of proof, if necessary, to take into account products using AI and ADM.
- Proportionality Test Directive (Directive (EU) 2018/958) — the resolution stresses the importance of properly assessing the risks before automating professional services.
We anticipate further clarity on the European approach to AI regulation in the coming weeks and months. Stay tuned for further updates.