The reference platform
Agent Liability Europe is an independent publication tracking the operator obligations created by Regulation (EU) 2024/1689, the AI Act, and by the revised Product Liability Directive. Its readers include in-house counsel, compliance officers, risk managers, chief technology officers and policy professionals in organisations that deploy autonomous AI agents in or into the European Union.
It is one of six sites published by Future Proof Intelligence on AI agent liability, certification, insurance and operator guidance. Organisations cited in our tracking, analysed in our briefings or included in our tools appear because their published positions matter to the field. Inclusion in our editorial map signals substance. It does not signal a commercial relationship.
Today: editorial only
This site runs on an editorial only basis. No carrier, vendor, law firm, consultancy, supervisory authority or other third party pays for inclusion, placement, ranking or favourable framing in any content here. Every source cited, every development tracked and every framework analysed qualifies on editorial grounds alone.
The commitment is set out in full in the editorial standards. Editorial and commercial work run on separate budgets and separate decision chains.
What qualifies for our tracking
These are the criteria for including a statute, guidance, court decision, supervisory opinion or framework in our tracking and analysis.
Primary legislation and regulations
An instrument qualifies if it creates, modifies or clarifies the obligations of operators or providers of AI systems in the European Union, including obligations under Regulation (EU) 2024/1689, Directive (EU) 2024/2853 (the revised Product Liability Directive), Directive (EU) 2022/2557 (the Critical Entities Resilience Directive) and related instruments. It must be published in the Official Journal of the European Union or the equivalent official record of the Member State concerned.
Supervisory guidance and opinions
A supervisory document qualifies if it is issued by a competent authority under the AI Act (a market surveillance authority, another designated national authority, or the AI Office), by EIOPA, EBA or ESMA, or by a national financial supervisor, and it sets or clarifies the conduct expected of operators in deploying, overseeing or reporting on AI systems. It must be published on the issuing authority's own website.
Court decisions and enforcement actions
A court decision qualifies if it sets or clarifies a principle relevant to the allocation of liability for AI agent conduct. This includes decisions of the Court of Justice of the European Union, the General Court, national courts of Member States, and the European Court of Human Rights where the decision bears on AI. Enforcement actions by data protection authorities, market surveillance authorities or national AI supervisors qualify when they produce a public decision or sanction.
Standards and technical specifications
A standard qualifies if it is published by ISO, IEC, CEN, CENELEC, ETSI or NIST and sets requirements or guidance that EU legislation or supervisory guidance for AI operators refers to directly. It must be in its published, official form.
A future partner tier
We may introduce a Partners tier for verified institutional participants that meet criteria published in advance. It has not launched, and nothing here is a commitment or an offer. The outline below exists so that interested institutions can see the intended direction and take part in the consultation that would come before any launch.
- Reference listing. A verified supervisory authority, law firm, consultancy or implementation partner with documented AI Act expertise receives a named profile, with editorial coverage of its published positions and guidance.
- Sponsored research briefing. A briefing on a topic the partner proposes, written by the editors, labelled as sponsored and published apart from editorial coverage, no more than one per quarter per site.
- Institutional partnership. A multi year arrangement across the sites, which may include licensing of the tracker datasets and co-branded research.
Every paid placement would be labelled as commercial. Partnership would not influence editorial coverage, tracking decisions or the framing of any analysis. A firm in the programme is not guaranteed coverage of its work, and a firm outside it is not excluded. Terms would be discussed directly with each institution.
What inclusion means
Being cited in our tracking, analysed in our briefings or named in our tools is not an endorsement by Agent Liability Europe or Future Proof Intelligence of any product, service, legal position or interpretation. We are not a regulator, a law firm or an accreditation body. We track and analyse published material; we do not certify compliance, validate legal positions or approve commercial conduct.
Citations in our articles are documentary references to published primary sources. They do not endorse the positions taken in those sources.
How to propose a source
We review publicly available primary sources continuously. If a statute, guidance document, court decision, enforcement action or framework that meets the criteria above is missing from our coverage, write to the editors with a direct link to the primary source and a short note on why it matters. Inclusion is editorial and free of charge, and a submission does not guarantee it.
Expressing interest in the partner tier
If you represent a supervisory authority, law firm, consultancy or implementation partner with a material interest in EU AI Act operator compliance and would like to hear if the partner tier opens, write to us. We will contact you when a consultation begins. We do not commit to a launch date.
Conflicts of interest
The conflicts that apply to this site are set out in section 5 of the editorial standards. Future Proof Intelligence is the publisher. It has a commercial interest in the development of the AI agent compliance market through its certification and intelligence work. That interest is disclosed and does not affect coverage decisions.