Von der Leyen’s centrist allies clash over AI liability rules - The Next Web

The European Parliament has held a plenary debate on whether AI companies should answer for the harm their models cause, which is a bit like a pub quiz on whether the landlord should answer for the beer. Socialist and Green MEPs want a shiny new AI liability act; centre-right MEP Arba Kokalari has looked at the AI Act and declared the rulebook already full. Full, note, not in force. Parts of it are still waiting for their moment, like a fire extinguisher still in its packaging while the kitchen smoulders.

This is the centrist alliance of Von der Leyen’s Europe in its natural habitat: a plenary debate about whether to add a page to a document that has not finished being a document. The Commission proposed AI liability rules as a directive back in 2022, which in Brussels arithmetic means the idea has now had a full childhood and is ready to start school. Meanwhile the actual question — should a company answer when its model maims someone’s credit rating, liberty or livelihood — is treated as a drafting matter. One can almost hear the white smoke being scheduled for a future trilogue.

Kokalari’s line is that the EU already has a full rulebook in the AI Act, parts of which are not yet in force. That is the legislative equivalent of saying the fire brigade is fully staffed, they just haven’t put on their boots. The Socialists and Greens asking for a liability act are not radicals; they are people noticing the gap between a press release and a court summons. But the machines will not wait for the AI Act’s commencement orders. They are already out there, causing harm in ways that will require a lawyer to describe. The EU’s answer is a debate about which document should eventually contain the paragraph that might, one day, mention it.