Share This Article
Just a few days after the provisional agreement on the EU AI Act, policymakers reached a landmark agreement updating the Product Liability Directive which now also covers liability deriving from artificial intelligence.
Here’s a quick summary of the key points:
Innovative Scope: The directive now includes digital products, such as software and AI technologies 📌 We shall see whether this means the end of the proposal of an AI liability directive
Liability Expansion: Manufacturers are now liable for defects in any component under their control – tangible, intangible, or related services 📌 The tricky aspect would be the definition of what falls under the control of manufacturers
AI and Machine Learning: A crucial addition is the treatment of products capable of learning and evolving, specifically targeting artificial intelligence based on machine learning techniques 📌 It shall be reminded that the AI Act does not cover aspects relating to AI liability, but just the compliance of AI systems. As such the two pieces of legislation now complement each other
Broadened Damage Definition: Damages under the directive now encompass death, personal injury, psychological harm, destruction of property, and significantly, loss or corruption of non-professional data 📌 This broader definition is the result of the evolution of our society
Evidence Disclosure: A pivotal change is the introduction of balanced evidence disclosure rules for both claimants and defendants, with protective measures for confidential information 📌 The purpose is to adequately protect the parties involved, but there is a risk of abuses.
Burden of Proof: The directive outlines specific conditions where product defectiveness is presumed, placing the onus on the defendant to disprove it. This includes scenarios of non-disclosure of relevant evidence or obvious product malfunction 📌 This approach might lead to a potential unbalance between the parties that might be risk in a scenario of technological evolution
Open Source Software: The directive does not apply to free and open-source software unless it’s commercially distributed 📌 This approach is in line with what agreed under the AI Act.
Compensation Fund: The EU may implement national compensation schemes for victims of defective products, particularly when the responsible entity is insolvent or nonexistent 📌 This is one of the few options viable to avoid that innovation is hindered by the lack of trust.
Liability Exemption: Small software manufacturers get a liability exemption, a boon for micro and small enterprises, provided certain conditions are met 📌 Also this provision is in line with what agreed under the AI Act where lower sanctions are provided for small entities and start ups.
Right of Recourse: In cases of multiple liable parties, each has the right to seek compensation from the others 📌 Probably, this solution was adopted since in a digital environment, the potential manufacturers are a multitude of entities.
Timeline & Implementation: Unlike the AI Act that is a regulation with direct effect in the EU Member States, the new product liability directive will be applicable to all EU market products 24 months post-enactment, with member states required to transpose it into national law 📌 Hopefully there will be consistency in the approach followed by the different EU member States.
This directive marks a significant step in aligning product liability with the digital age, reflecting the complexities of modern tech products and services. The EU wants to take a critical step in innovation, and that step passes also through new regulations and more certainty.
If you missed the latest updates regarding the provisional agreement on the AI Act, the following article may be of interest “EU AI Act Approved: Everything You Need to Know on the Artificial Intelligence legislation in Europe“.