MEPs push for improvement in AI legislation to protect consumers
About this publication
The lead committees in the European Parliament want beefed-up protections for consumers in the AI Act, which would be the world’s first law addressing artificial intelligence. The text approved today in the civil liberties (LIBE) and internal market and consumer protection (IMCO) committees would clearly improve the Commission’s proposal. It addresses crucial issues that the Member States ignored in their position adopted in December last year.
Ursula Pachl, Deputy Director General at the European Consumer Organisation (BEUC), said:
“With the explosion of ChatGPT and a daily stream of news about the latest AI systems, it is clear we are only at the beginning of the AI age. Although AI may improve our lives in many ways, there are well-founded concerns that AI systems can also harm consumers. People must be properly protected against the risks of these new technologies.
“The outcome of today’s vote means that the European Parliament is determined to improve consumer protection and respect the fundamental rights of users of AI systems. Consumers can breathe a sigh of relief given the disappointing proposal of the Commission and position of Member States so far.
“For instance, proposed bans on the use of facial recognition in publicly accessible spaces or social scoring by private entities are big steps forward. It is more than welcome that MEPs want to grant consumers specific individual rights. These include a right to be informed when a consumer is subject to a decision from a high-risk AI system, for example when requesting a loan or insurance policy, a right to complain, or a right to join a collective redress action. In the case of ChatGPT, we welcome the Parliament’s resolve to regulate large language models. But it remains to be seen if this will be enough to properly protect consumers.
“We are however very worried that AI operators could circumvent the rules of the AI Act by simply saying their services are not high risk. Also, the proposed basic principles for “trustworthy AI” which importantly would apply to all AI are only voluntary for businesses or public entities, when they should be obligatory.”
Outcome of the vote
MEPs in the LIBE and IMCO committees voted to:
-
Ban remote biometric identification, such as facial recognition, used by private entities in publicly accessible spaces. This is a positive step as this technology is too invasive and undermines our fundamental rights. The Commission had already banned it when used by public authorities in public spaces.
-
Ban social scoring when used by private entities, which is when an AI system evaluates the trustworthiness of an individual based on their social behaviour or their preferences, emotions, health or intelligence. It is too invasive and too arbitrary a practice to be used on consumers.
-
Give consumers the right to ask for collective redress when an AI system has caused harm to a group of consumers.
-
Consumers are granted several rights, including a right to be informed when being subject to a decision from a high-risk AI system, a right to complain to an authority about an AI system, the right to bring a supervisory authority to court if it fails to take action.
-
Subject generative AI such as ChatGPT to specific rules, such as obliging AI developers to identify, reduce and mitigate the risks to health, safety and fundamental rights before placing the AI system on the market.
-
However, there are also some significant negative aspects to today’s votes:
-
The two committees have blurred the lines of what constitutes a high-risk AI system which would give developers too much discretion to decide if their system is considered high-risk or not. The committees have not pushed for third party certification about whether the system conforms to the AI Act or not.
-
The principles which BEUC called for to apply to all AI regardless of the level of risk are only voluntary, which will limit their usefulness.
-
No ban on emotion recognition when it is used on consumers, which can lead to serious intrusions in consumers’ privacy and ability to take autonomous decisions
-
The next step is for the Parliament to vote in plenary in June. Then negotiations between the Parliament and Member States can start to determine the final text. The upcoming Spanish Presidency wants to reach a political agreement by the end of the year.
Notes
For further information, you can consult BEUC’s factsheet on AI and our checklist to EU decision-makers.