EU lawmakers clash over reviving AI liability rules

European Parliament lawmakers are divided over whether to revive legislation holding AI companies liable for harm caused by their models. Socialists and Greens urged the European Commission to restore its proposed liability directive, while centre-right lawmakers argued that the existing AI Act provides sufficient regulation. The Commission proposed the directive in 2022 and withdrew it last year during a push to simplify EU law and improve competitiveness.

Alex Agius Saliba, vice-president of the Socialists and Democrats, called the withdrawal a mistake. Dutch Green MEP Kim van Sparrentak said AI developers should face responsibility comparable to manufacturers of faulty products, with liability extending to testing and development. OpenAI chief executive Sam Altman has also floated a framework covering companies whose models can independently conduct cyberattacks. The Commission is investigating OpenAI and Anthropic over agent hacks.

Swedish centre-right MEP Arba Kokalari countered that parts of the AI Act have yet to take effect and urged less bureaucracy, more private investment and better access to data and computing infrastructure. Kokalari led Parliament’s work on a package that weakened the AI Act and delayed a core provision. Commissioner Jozef Síkela said the liability proposal was withdrawn because EU countries and lawmakers could not agree on it.