The EU AI Act enables users to lodge a complaint with a market surveillance authority. A market surveillance authority is to be set up by each Member State in accordance with Article 10 of the Regulation (EU) 2019/1020. The regulation enables market surveillance to improve the functioning of EU’s market by providing protection of public interests such as health and safety of the customers, protection of the environment and public security.
-
What does the provision entail
Enables any natural or legal person to raise a complaint. The administrative and judicial remedies remain intact with the person.
Complaints must be investigated in accordance with the procedure set out by the Regulation (EU) 2019/1020.
-
AI product liability
The right to lodge a complaint appears to be an attempt at fixing product liability for AI systems. AI systems providers (including deployers and entities putting them on the market) have relied on treatment of AI as a distinct technology incapable of being defined within current legal context. However, the inclusion of a right to lodge a complaint with the market authorities treats AI products on par with other manufactured products or services.
-
Judicial scrutiny of AI products
This comes on the back of judicial scrutiny of AI-based products and false advertising claims. The Federal Court in US state of Florida recently declined to entertain an appeal by Tesla from a case that fixed liability on the company for defective self-driving leading to a crash. A lower court had found Tesla 33% liable and ordered a penalty of $243 million on Tesla. The accident occurred as the owner of an autopilot-equipped Model S drove through an intersection into a car parked at the shoulder of the road, killing a woman and injuring her boyfriend. Liability was fixed on Tesla despite Tesla’s claims that the driver was recklessly driving their car. Back in 2020, a Dutch court had ruled the digital welfare fraud detection system called Systeem Risico Indicatie (SyRI) as unlawful. The Court held that the system failed to conform with data privacy regulations. These cases indicate a willingness of the judicial system to treat AI on par with any other product.
-
Necessity of express AI-product liability
Explicitly providing for AI liability is a welcome step.
It is particularly important in jurisdictions with lax regulations such as India where no express liability arises for AI systems. In such cases, the traditional understanding of liability is unlikely to cover AI products. This may lead to gaps which judicial interpretation alone is unlikely to fill at the pace which AI-based technologies develop and are put on the market.