EventsPodcasts
Find Us
ADVERTISEMENT

Incoming European Parliament to work on AI liability rules, despite tech lobby concerns

An LG CLOi CoBot is configured to wash dishes in a mock restaurant.
An LG CLOi CoBot is configured to wash dishes in a mock restaurant. Copyright John Locher/Copyright 2020 The AP. All rights reserved.
Copyright John Locher/Copyright 2020 The AP. All rights reserved.
By Cynthia Kroet
Published on Updated
Share this articleComments
Share this articleClose Button

Consumer groups support the new law to protect people from faults caused by artificial intelligence (AI).

ADVERTISEMENT

The incoming European Parliament will continue work on a legal framework to determine who is liable in case of defaults in artificial intelligence products, now that the AI Act will take effect on 1 August. But the Brussels tech lobby and consumer organisations remain divided over the need for additional rules. 

The European Commission proposed the AI Liability Directive in 2022 in a bid to modernise existing liability rules with new provisions covering harms caused by AI systems to ensure uniformity of protection. 

The file didn't move forward within the Parliament during the last mandate: lawmakers decided to put it on hold until an agreement was reached on the AI Act - the main framework which regulates AI systems according to a risk-based approach.

The AI Act itself, the world’s first stringent rules to regulate high-risk machine learning systems, will officially enter into force next month. The general-purpose AI rules will apply one year after entry into force and the obligations for high-risk systems in three years' time.

The lawmaker in charge of steering the AI Liability Directive through Parliament is Axel Voss (Germany/EPP). Voss, re-elected in the June EU election, told Euronews that “it would be better to have an AI liability regime in place.” 

The tech lobby in Brussels worries however, that liability rules specific to AI will cause additional regulatory hurdles for companies. CCIA Europe, an organisation representing IT and telecom companies, said it is “strongly against adding unnecessary and burdensome rules on companies trying to compete in an already highly regulated market.”

They claim that the issues are already covered under the revamped Product Liability Directive (PLD): rules adopted in 2023 by the EU Parliament and member states, to replace a set of 40- year-old framework, which takes into account technological developments. 

CCIA Europe’s Senior Policy Manager Boniface de Champris said that with the recent adoption of the AI Act and the PLD, “the introduction of additional AI liability rules is highly questionable and likely unnecessary.”  

These comments were echoed by Thomas Boué, Director General of Policy EMEA at BSA, The Software Alliance.  “The proposal is relatively short and overlaps significantly with the recently adopted Product Liability Directive. The first step should be for the European Parliament and Council to assess whether this text is still necessary,” he said.

Redress

Consumer groups, however, still think there are gaps in the current legal framework. Agustín Reyna, the new Director-General of consumer group BEUC, told Euronews that it’s “the missing piece of the puzzle.”

“It's about what is happening if something goes wrong and if a product causes harm. We have rules for product safety, so it makes sense to also look at this through the lens of the risk posed by these new technologies,” Reyna said. 

Els Bruggeman, head of policy and enforcement at Euroconsumers, a pan-European consumer trade group, said that for her member organisations too, this is a priority file. For the AI Liability Directive to be effective, the organisation is asking lawmakers to avoid a fault-based approach.

 “This would make it very hard, if not impossible, for consumers to get redress. After all, how will consumers ever be able to prove the fault? Even if they do get access to the data and algorithms, they’ll almost need a degree in engineering to ever be able to understand and demonstrate how they work,” Bruggeman said.

“We’d like legislators to go one step further and introduce a reversal of the burden of proof: consumers should only have to prove the damage they suffered and the involvement of an AI system. And we also need the Directive to include all types of automated decision making - not only fully automated - and all kinds of harm," she added.

Between the 27 EU member states, preliminary discussions have been held on working party-level. Not much progress is expected before the end of this year: it has not been mentioned as a priority file for Hungary, the country chairing the meetings between the governments in the second half of this year.

Share this articleComments

You might also like