To ensure that Artificial Intelligence is used responsibly and safely in the EU, the European Commission presented the draft regulation to regulate the use of Artificial Intelligence on 21 April 2021. On 11 May 2023, after almost a year of difficult negotiations and around 4,000 amendments tabled, MEPs from the lead committees IMCO und LIBE agreed on a joint proposal for an AI Act.

More rights for consumers when using AI
From the point of view of AK, it is basically to be welcomed that a regulatory framework for AI is being created by the EU. AI is influencing more and more areas of consumers' daily lives, such as creditworthiness checks, mobility or micro-targeting in online advertising. While the original proposal of the EU Commission showed immense deficits in the area of consumer protection, it is extremely welcome that the EU Parliament now wants to improve consumer protection. “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”, said Ursula Pachl, Deputy Director General of BEUC, the European Consumer Organisation.

In particular, the EU Parliament would like to anchor an explicit ban on the use of biometric identification methods in real time in public spaces or of social scoring – the latter is also provided for in the Council agreement. Particularly positive are also the rights granted to consumers, including the right to collective redress and notification and explanation obligations of the system operators.  The EU Parliament would like to introduce its own rules for systems such as ChatGPT. However, the EU Parliament has not yet taken up all concerns that were expressed by AK. Emotion recognition, for example, should not be included in the list of “prohibited practices”. Another criticism is that developers are given too much discretion to decide if their system is considered high-risk or not, and there is no mandatory third party certification about this information. In addition, principles such as transparency, fairness and accountability, which are demanded by consumer protection organisations  on a risk-independent basis for AI systems are still voluntary, which diminishes their usefulness.

Special protection needed for the use of AI in the workplace
The AI Act does not go far enough in the world of work, the issue of artificial intelligence at the workplace remains largely unconsidered and the necessary protective provisions are therefore lacking. AK demands a strict regulation of AI applications that affect workers' rights, working conditions and health at work. The AI Act must therefore enable the Member States to retain existing labour (constitutional) law regulations and to regulate the use of AI at the workplace with national regulations (collective agreements). Last but not least, it is urgently necessary to create a separate EU Directive on Artificial Intelligence in the workplace with minimum protection provisions for workers.

On 14 June 2023, the vote in the plenary of the EU Parliament will follow, before the trilogue negotiations with the Council can start, which has already defined its position on the AI law at the end of 2022. It is planned to start the negotiations under the Swedish Council Presidency; the incoming Spanish Council Presidency has announced that it will strive for a political agreement by the end of the year.

Further information:
AK EUROPA Position Paper: AI Liability Directive
AK EUROPA: New AI liability proposal doesn’t look promising
AK EUROPA: Open Letter of AK to the Members of the EU Parliament on the need for improvement to the draft Artificial Intelligence Act from the consumer's point of view 
European Consumer Organisation (BEUC): Factsheet Artificial Intelligence