Last updated: August 2026
RegDossier is researched and written with AI, working from primary regulatory documents, and every claim links to the source it came from. No lawyer reviews these pages, and nothing here is legal advice. This page sets out what that method covers and where it stops, because a publication that reports other people’s transparency obligations cannot be vague about its own.
AI Does the Reading Because Nobody Else Could
Eight frameworks across twenty-seven member states produce guidance, delegated acts, transposition laws, enforcement decisions and court referrals faster than any small publication follows by hand. AI reads the Official Journal and EUR-Lex, Commission and Council publications, ESMA, EBA, EIOPA and ENISA guidance and national regulator announcements, then drafts the page. That volume is the entire reason the method is what it is. The alternative was one regulation covered slowly, not eight covered by people. What that choice costs is set out immediately below, in the same detail.
Nobody Reviews This Before It Goes Up
There is no editor reading behind the AI, no lawyer signing anything off, and no second pair of eyes on the numbers. Saying otherwise would be the easiest sentence on this page to write and the least defensible. What does happen before publication is narrower: someone opens the links and confirms each one lands on the body it claims to. That checks the address. It checks nothing about the content, which is precisely why the pages are built the way they are described below.
The Directory Lists Firms, It Does Not Rate Them
Our provider directory names firms that publish a service for a given regulation on their own website, with a link to that page and the date the link was last opened. It carries no ratings, no stars and no rankings, because there is no process here that could produce an honest one.
Inclusion means the firm sells that service and says so publicly. It does not mean we vouch for the work. A firm gets in because it does the work, never because it paid, and where a listing is promoted the listing itself says so. Choosing a provider, and the due diligence that goes with it, remains yours.
The Link Next to the Claim Is the Point
Dates, article numbers, turnover thresholds and penalty tiers are what language models get wrong, and they get them wrong with complete confidence. So the source sits next to the claim instead of in a list at the bottom, and the position we take is that you should use it. Where our text and the Official Journal disagree, the Official Journal governs. Treat RegDossier as an index to the regulation and a way of working out which forty pages belong to you, not as a substitute for opening them.
What This Method Does Not Do
It does not give you a compliance opinion. It does not tell you whether a regulation reaches your organisation. It does not tell you how a national authority will read an ambiguous provision, and new EU law is full of those. It does not replace counsel on anything that carries a deadline or a penalty for you. Where sources disagree, we would rather show you both than pick the tidier one, and where a date depends on a text that is signed but not yet published, we say so and mark it.
Every Tool Threshold Traces to a Named Article
The checkers, calculators and assessors run fixed logic in your browser. Every threshold in them is taken from a specific provision and cited in the code, and where we cannot point at the article, the calculation stays out of the tool. Each one shows the date it was last updated. A tool tells you where you probably sit and what to read next, which is a long way from telling you that you comply.
How Often Anything Here Gets Re-Checked
The register is reviewed monthly. That review opens every link, catches authorities that have been renamed or merged, and picks up member states that have designated a body since the last pass. Each row shows the date its links were last opened, so you can see how fresh it is. Tools are audited in full twice a year and again whenever a monthly review turns up a change that touches one. Amending regulations move dates and thresholds in a way that ripples through several pages at once, and that is the moment things go stale. It has happened here. When it does, the fix and its date go on the page.
Corrections Are Dated and Left on the Page
Send errors to info@regdossier.eu and we fix the page, date the correction and leave a note saying what changed, including on articles old enough that nobody would go back and check. Quietly editing a wrong date to look right is how a regulatory publication stops being one.
The Underlying Data Is Yours to Take
The trackers are published under a Creative Commons Attribution 4.0 licence, with the tables downloadable as CSV and JSON. Credit RegDossier, link to the source page, and the figure is yours to put in a client alert, a slide or your own product. Publishing the raw files is part of the method rather than a courtesy. A dataset anyone can download and check against the gazette is a stronger claim to accuracy than any assurance we could write on this page.
Why This Page Exists Under the AI Act
Article 50 of Regulation (EU) 2024/1689 obliges deployers of AI systems that generate or manipulate text published to inform the public on matters of public interest to disclose that the content is artificially generated. Regulatory analysis sits squarely inside that, and the disclosure appears on every page. An exemption exists where a named natural or legal person holds editorial responsibility for the content, and we have not claimed it, because claiming an editorial process you cannot evidence is a worse position than describing the one you actually run. This page goes further than the disclosure requires, on the view that readers who came here for a straight answer about EU law can take one about us.