Everything on the Monitor is built to be checked. What the labels on an item mean, how an item gets onto the site and how the index is scored are all set out below, so that a reader who disagrees can find the exact point where we part company.
What the labels on an item mean
Every news item is a short summary of one named source, and a few labels sit around it. The two that appear on every item are short tags: V or R for the grade, and AI or AI+E for how it was drafted. HS is kept for a piece a person wrote, and no news or opinion item carries it. On a computer, hold the pointer over a tag to see the word in full; the key on every item page and at the foot of every page spells them out.
- Verified (V) means the primary document (the law, the contract, the official figure) was opened and read at the passage that says what the item says. That shows the source says it; it does not prove the source is right. Reported (R) means the primary document could not be reached, so the item rests on named reporting and should be treated as weaker.
- What the source does not establish says where the claim stops. A cloud service could reach the top assurance level in the proposed Cloud and AI Development Act and still run entirely on imported chips, because the levels never look at the chips.
- Index score is the item's score in the index: which way it moves Europe, and by how much. How the index works is explained below.
- AI-drafted (AI) or Editor-read (AI+E) says whether a person read the item before it went live. An AI system drafts every news and opinion item; AI-drafted means it went live without a person reading it, and Editor-read means the editor read it first, on the date the item gives.
- Human-written (HS, for Homo sapiens) is the tag for a piece a person wrote. No news or opinion item carries it, because an AI system drafts all of them.
- Opinion marks a position someone argues rather than something that happened, and an opinion never moves the index. Abroad marks an item whose actor or author sits outside Europe, which here means the EU, the EEA, the UK and Switzerland.
- Still open names a question we will check again, and the date we will check it.
- Correction shows what was changed and when, with the original wording quoted.
Grant figures from CORDIS, the Commission's own register of signed grant agreements, count as Verified. They show money contracted, not money paid, and two organisations in one consortium are co-members, not partners.
The four questions
Sovereignty is at least four questions, and the answers often differ: who owns it, who operates it, whose law governs it, and who made the parts. Most European rules test the first three and stop before the fourth, and that gap is much of what this publication tracks. When a sovereignty claim sounds strong, the useful question is which of the four it answers.
How the record is made
An AI system, Claude (Anthropic), drafts every news and opinion item from the one source it cites: the headline, the summary, the "does not establish" line, the grade and the index score. Before the item goes live, the pipeline checks it by machine. It fetches the source again and looks for the figures and names the item uses, a second model reads the item against the source, and the record is searched for a duplicate. An item that passes goes live, labelled AI-drafted. An item that fails is held, and the editor either releases it or drops it.
The editor reads every held item, every correction a reader sends, and the monthly issue, which is written by hand. In the archive months being re-read, the editor read every Reported item and every item the checks flagged from June 2025 onwards; from May 2025 back, those items went live without the editor reading them and carry AI-drafted, and an item that failed the checks there was dropped without being read. The weekly email is assembled from items already published and adds no new text. The editor does not read every daily item before it goes live, and that is why each item says whether it was AI-drafted or Editor-read. This is the item-by-item disclosure that Article 50(4) of the EU AI Act asks of a publisher, and the same field travels with the open data.
Beyond the summary, the grade and the score, the AI system writes no argument of its own, so the analysis is the reader's. A wrong grade or score found after publication is fixed as a correction. Editorial responsibility rests with RB Consult AB, Reidar Balstad, editor.
The index
The European Sovereignty Index gives two numbers, for Europe and for each of four areas: compute and infrastructure, policy and law, defence and space, and industry, investment and R&D. (The areas are topics; they are not the four questions above.)
The first number shows where Europe stands, on a scale of 0 to 100. It is the average of twenty measurements, five per area, such as Europe's share of the world's chip production. When a measurement has a goal, its score shows how close Europe is to it: Europe makes 9 percent of the world's chips against the EU's goal of 20 percent, which scores 45. The targets behind the index lists all twenty with their goals and sources. The measurements are checked about once a quarter, and a score of 100 means a goal is met, not that Europe is sovereign.
The second number shows which way this year's news has pushed Europe. Every news item from 1 January 2026 gets a score, shown on the item under "Index score", and the score is built from five things:
- the direction: towards autonomy, towards dependence, both ways, or no change;
- how many of the four questions the item changes, from 1 to 4;
- how binding it is: announced (1), agreed or funded (2), or in force and operating (3);
- how big it is: small (1), medium (2) or large (3), with the lines at EUR 100 million and EUR 1 billion, or at EUR 500 million and EUR 5 billion in defence;
- how sure we are: Verified counts in full and Reported counts half.
Items dated before 1 January 2026 are also in the record, but they do not count in the index, and each one says so on its own page.
The score is those multiplied together, so the biggest possible item scores 4 x 3 x 3 = 36, which counts as 1.0 on the charts. For example, the EU's agreement letting Canada into SAFE defence buying moves one question (who made the parts) towards dependence, is agreed (2), large (3) and Reported (half): 1 x 2 x 3 x 0.5 = 3 of 36, or minus 0.08. Each month adds up its items, autonomy minus dependence, and the running total starts on 1 January 2026. An item that moves both ways is scored as two parts, one in each direction.
Both numbers combine the four areas with the same weights, by default compute 30, policy 25, defence 25 and industry 20, and you can change them on the index page. Two rules guide the scoring: running a service in Europe is not sovereignty when the owner and the governing law stay outside it, and the EU's own targets are the yardstick wherever they exist. The news never changes the first number, and because our picks lean towards European moves, the second number leans positive.
The monthly issue's markers
The monthly issue marks each development with one of four markers, an older and simpler count than the index, and the two are never added together. The markers are ▲ towards autonomy, ▼ towards dependence, ◆ both directions, and ■ no movement, which covers what is announced or proposed but not yet real. In the index, by contrast, an announcement still counts, at the lowest weight.
A section's net is its ▲ count minus its ▼ count. In issue zero, twenty-nine developments netted plus eight, but policy and law ran at minus one while defence and space ran at plus five. That is why every section carries its own count and the total is never quoted alone.
Corrections
Send the primary document and the provision to info@sovereigntymonitor.eu, or reply to any email. If it says what you say it says, the item changes and your name goes on the correction unless you ask otherwise. A disagreement about interpretation may become an added line rather than a correction, because two people can read one text honestly and differently. A Reported item is provisional by design, and a reader with access to the primary text is the most common way it becomes Verified.
What this is not
It is not investment advice, it is not a vendor evaluation, and it is not a ranking of countries or firms. Suppliers appear when a rule or a claim makes them relevant, quoted in their own words, and the scores rate developments one at a time, never anybody's standing.