Directive on adapting non-contractual civil liability rules to artificial intelligence
European Union · Regulation · European Commission
Our summary
A proposed directive to adapt civil liability rules so that victims harmed by artificial intelligence systems can obtain compensation more easily. It addresses challenges that existing fault-based liability rules pose for AI-related damages by harmonising rules across the EU to reduce legal uncertainty and fragmentation. The proposal aims to promote trust in AI and its uptake by ensuring victims receive protection equivalent to those harmed by other products.
Written by the Index in our own words. It is not legal advice; the official text is the authority.
Inside this policy
Every section of this policy in order. Our notes summarise each one in plain language and do not quote the official text, which is the authority.
Drafted by the Index An automated summary of the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
9 sections
Jump to a chapter
Article 1 Subject matter and scope
Establishes common rules on disclosure of evidence for high-risk AI systems in fault-based damage claims and on burden of proof in such claims brought to national courts. Applies to non-contractual fault-based damage claims arising from AI systems after the transposition deadline, but excludes criminal liability and transport, product liability, digital services and certain national proof rules.
Article 2 Definitions
Defines AI system, high-risk AI system, provider, and user by reference to the AI Act. Defines claim for damages as seeking compensation for AI output or failure to produce required output, and claimant as an injured person, successor, or representative bringing such a claim. Also defines potential claimant, defendant, and duty of care.
Article 3 Disclosure of evidence and rebuttable presumption of non-compliance
Empowers national courts to order providers, obligated persons, or users to disclose evidence of high-risk AI system malfunction after the claimant has made proportionate attempts to gather it from the defendant. Courts must limit disclosure to what is necessary and proportionate, may restrict confidential information, and must presume fault if the defendant fails to comply with disclosure orders.
Article 4 Rebuttable presumption of a causal link in the case of fault
National courts shall presume causal link between fault and AI output or failure where the claimant demonstrates fault, it is reasonably likely the fault influenced the output, and the output caused damage. For high-risk AI systems, fault of the provider is presumed when they failed to meet requirements on training data quality, transparency, human oversight, accuracy or cybersecurity, or failed to take corrective action, and fault of users is presumed where they failed to comply with use, monitoring, suspension instructions or exposed the system to irrelevant input data. The presumption does not apply if the defendant shows sufficient evidence and expertise are reasonably accessible to the claimant, does not apply to non-high-risk systems unless proving causal link is excessively difficult, and does not apply to personal non-professional users unless they materially interfered with operation.
Article 5 Evaluation and targeted review
Article 6 Amendment to Directive (EU) 2020/1828
Adds this AI Liability Directive to Annex I of Directive (EU) 2020/1828.
Article 7 Transposition
Member States must enact implementing laws within two years of entry into force and communicate them to the Commission, referencing this Directive or making such reference at official publication.
Article 8 Entry into force
Article 9 Addressees
This Directive is directed to the Member States. The text includes signatures from the European Parliament and Council.
Controls
What this policy requires, forbids, grants or penalises, one control at a time and in our own words. Each control links to the official text, which is the authority. This is not legal advice.
Drafted by the Index Drafted automatically from the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
Themes covered 0 of 35
Choose a theme to see only its controls. About the themes
Risk and scope 0 of 4
- Risk classification, not covered
- Prohibited practices, not covered
- Risk management system, not covered
- Impact assessment, not covered
Data and privacy 0 of 4
- Data governance and quality, not covered
- Data subject rights and consent, not covered
- Cross-border data transfer, not covered
- Copyright and training data, not covered
Transparency and records 0 of 4
- Transparency and disclosure to users, not covered
- Labelling of synthetic content, not covered
- Technical documentation, not covered
- Record keeping and logging, not covered
Technical safeguards 0 of 4
- Human oversight, not covered
- Accuracy, robustness and cybersecurity, not covered
- Bias and non-discrimination, not covered
- Testing and evaluation, not covered
Assurance and lifecycle 0 of 4
- Conformity assessment and certification, not covered
- Registration and notification, not covered
- Incident and breach reporting, not covered
- Post-market monitoring and corrective action, not covered
Rights and protected groups 0 of 4
- Automated decision-making rights, not covered
- Children and vulnerable groups, not covered
- Biometric identification and surveillance, not covered
- Whistleblowing, complaints and redress, not covered
Roles and organisation 0 of 6
- Supply chain and provider-deployer duties, not covered
- General purpose and foundation models, not covered
- Accountability and governance roles, not covered
- Training and AI literacy, not covered
- Public sector use and procurement, not covered
- Innovation support and sandboxes, not covered
Enforcement and sectors 0 of 5
- Enforcement, penalties and regulators, not covered
- Sector rules: health, not covered
- Sector rules: finance and insurance, not covered
- Sector rules: employment, not covered
- Sector rules: education, not covered
The full list of controls could not be loaded, so only the first 0 are shown.
Change history
- Official text wordingSource for this change (opens in a new tab)
- Official text wordingSource for this change (opens in a new tab)