EU Directive 2024/2853: Liabilities for companies selling AI agents
In brief
- Directive 2024/2853 classifies AI agents as products, including SaaS.
- Companies face strict liability for physical, psychological harm or non-professional data destruction caused by a defect.
- A substantial update after December 2026 subjects an existing agent to the regime.
The EU Directive 2024/2853, to be transposed by December 9, 2026, classifies AI agents as products, including SaaS. According to the announcement, companies developing or selling these agents under their brand become strictly liable for physical harm, medically recognized psychological damage, or non-professional data destruction caused by a defect. A substantial update after this date subjects an existing agent to the regime. In France, no transposition text was identified as of October 20, and a January draft by the Ministry of Justice omitted three points according to the Paris Bar.
This directive extends manufacturer liability to AI, with a presumption of defect if evidence is not disclosed. Financial losses remain outside the scope, as noted in JPMorgan Chase's example. Companies must adapt processes to comply with these new rules or face penalties.
What it changes for you
- LegalLaw firms will need to advise clients on new liability obligations and legal risks associated with AI agents.
- Public SectorPublic administrations must transpose the directive by December 2026 and adapt regulatory frameworks for AI agents used or sold.
This summary gives you the essentials. The full article, its figures and quotes are on ActuIA.
Read the original articleAI-generated summary of the source, checked by rules. How we write · Report an error