Who is responsible when AI causes harm?
AI has no legal personality: responsibility always traces back to people — provider, deployer or user — depending on the case. The framework combines general liability law, the revised EU directive on liability for defective products (now covering software), and AI Act obligations.
An AI has no legal personality: it cannot be sued or held responsible. Responsibility always traces back to natural or legal persons, depending on their role: the provider who designed the system, the deployer who uses it, or the user, depending on the origin of the damage. The framework combines general liability law, revised European product liability law — which now covers software — and the obligations of the AI regulation. Nobody hides behind the machine.
The principle: human responsibility
No AI system is a legal subject. In the face of damage, the responsible person is sought. Depending on the case, this may be the designer for a system defect, the deployer for faulty use or a lack of supervision, the user for misuse.
Legal foundations
General civil liability law applies: fault, damage, causal link engage the liability of their author.
European product liability law for defective products has been revised to explicitly cover software and AI systems, facilitating compensation for victims of defective products, including digital ones.
The AI regulation imposes obligations on providers and deployers of high-risk systems, non-compliance with which may engage their liability.
Why it is evolving
The attribution of responsibility is becoming more complicated with the growing autonomy of systems, particularly agents. European law is adapting its frameworks so that victims do not encounter a void. The constant direction remains that responsibility lies with identifiable actors.
What you can do
In case of damage caused by an AI system, identify the actors involved, preserve evidence, and assert your rights according to the same principles as for any defective product or service, relying on applicable liability regimes.
Frequently Asked Questions
Can an AI be sued?
No. The responsible persons are sued: designer, deployer or user depending on the case.
Are software products covered by product liability?
Revised European law now explicitly covers software and AI systems.
Who provides compensation in case of damage?
The person whose liability is established, according to applicable regimes.
ICIA Resource
ICIA offers MentivisOS ICIA, a free lifelong platform to learn AI with a personalized learning path.
Discover MentivisOS ICIA