Summary
- RIPE NCC's sanctions reporting counts parties frozen, parties held, screening alerts and the resources inside those registrations, while omitting every case-level detail; the account of the report attributes that omission to privacy and legal constraints.
- The count is checkable at the aggregate and uncheckable at the case: a reader learns that a registry applied EU and U.S. sanctions to members and End Users, and gets nothing with which to test whether any individual measure was right.
The freeze is not of traffic. It is of the registry's own entry, and that distinction does more work than any figure in the report. CircleID's account of a RIPE NCC sanctions transparency report describes administrative freezes of internet number resources rather than service termination or routing changes. A member who loses a registration has not necessarily gone dark; it has lost the registry's word for it. That is a serious thing to lose, and it is a different thing from an outage. Fixing that difference first keeps the rest of this honest: nobody's packets stopped moving because of what follows.
RIPE NCC's own legal pages, as described in the material read on 17 November 2026, set out a formal sanctions procedure for complying with applicable sanctions and a set of transparency reports behind it (the sanctions procedure; the sanctions transparency reports). Neither page was read in its own words here: both came back as described readings, without a single quoted sentence, so nothing in this article is put in the registry's mouth. What the two descriptions do establish is the shape of the instrument — a procedure said to cover the identification of sanctioned entities, a decision-making process, notification requirements to members and available review or appeal mechanisms; and a reporting practice said to aggregate case data over defined reporting periods while generally omitting identifiable case detail, in the description's own words because of privacy and legal constraints.
Note what that last point is and is not. It is a description of a design decision, reported at second hand. It is not a finding about motive, and nothing here should be read as an accusation of concealment. A registry enforcing another jurisdiction's sanctions list has obvious reasons to avoid publishing who got caught by it. The interesting question is what such a design costs the reader, and the answer only becomes visible when the numbers are laid out by stage rather than piled together.
Six measurements, not one
The count itself comes from a single passage. Retained verbatim from the third-party account:
According to the Q4 2026 sanctions transparency report, the registry also placed four members on hold because their sanctions status could not be conclusively established. One of those members additionally holds resources as an End User. . . As of October 1, RIPE NCC’s cumulative screening records showed 2,339 alerts involving EU and U.S. sanctions checks. Of these, 2,108 had been resolved as false positives, inapplicable sanctions, exemptions or cases involving U.S. sanctions, while 119 remained under investigation and 39 had not yet been examined. Another 20 were classified as on hold.
Read slowly, that passage is not one number but six, and they measure different things.
The first is parties frozen: five members and 23 End Users, counted as two populations in the same reported action. The two together are 28 parties. That total is mine, a count of the account's own two figures, and it is not a population under sanction — the four members below are separate from it, and the account offers no cumulative party count at all.
The second is parties held: four members placed on hold because their status could not be conclusively established. Held is not cleared and held is not frozen. It is a third state, and the account treats it as one.
The third is the screening load: 2,339 alerts involving EU and U.S. sanctions checks, described as cumulative records, dated in the quoted words to October 1. The year 2026 around that date comes from the surrounding account, not from the quotation, and belongs to the account rather than to the registry's own sentence.
The fourth is what happened inside that load, and this is where the arithmetic starts to leak. Of the 2,339 alerts, 2,108 were resolved — but the sentence bundles four distinct reasons into that resolution: false positives, inapplicable sanctions, exemptions, and cases involving U.S. sanctions. A false positive is an error corrected. An exemption is a permission honoured. A U.S.-sanctions case is a different legal regime entirely.
Counting all four as one bucket means the published figure cannot be read as a count of mistakes, and anyone who does read it that way will overstate the error rate or understate it, depending on which reason dominates.
The fifth is work in progress, and it too is three states wearing one heading: 119 under investigation, 39 not yet examined, 20 classified as on hold. Investigation, queue and limbo are not the same condition, and only one of them involves anyone looking at the file.
The sixth is the resource inventory, and it arrives in three incompatible units: 4,850,688 IPv4 addresses, 29 IPv6 allocations and 82 autonomous system numbers held in frozen or on-hold registrations, per the same account's summary. Addresses, allocations and ASNs cannot be added, averaged or compared to each other. They also cannot be divided by any party count: nothing in the account says those resources belong to those 28 parties, or that both figures cover the same window. A per-party average here would be a number I invented, so there isn't one.
The 53 alerts the quoted sentences do not place
On the sentences quoted above, the four named outcomes account for 2,286 of the 2,339 alerts — 2,108 plus 119 plus 39 plus 20. The passage does not place the remaining 53 in a named category. That is my subtraction, performed on a retained quotation, and it says something narrow: an excerpt is partial by its construction, and a report may well place those alerts somewhere the excerpt does not reach. The quoted sentences simply stop before doing so. Anyone reading the same paragraph and assuming the total reconciles is making the same unstated assumption the rest of this article is about.
The identifier that was removed on purpose
Now put the six measurements back together and ask the only question a reader outside the registry can ask: which freeze was wrong?
The tally cannot say. It carries no case reference, no decision date, no category per case, no stated ground, and no record of what happened next. Every party appears as an integer. The description of the reporting practice says the omission of identifiable detail is deliberate, for stated privacy and legal constraints, and that the reports instead carry volume and categories of action. Deliberate is the word that matters. This is not an oversight to be fixed with a footnote; it is the specification.
That specification has a consequence the design may not intend. The described sanctions procedure includes notification requirements to members and available review or appeal mechanisms. So the instrument contemplates that a member is told, and that a member may contest. But the published tally shows neither: no visible notice, no visible review, no visible outcome of any review. Whether notices are given promptly, whether reviews are requested, whether any review has ever changed a freeze — none of it can be established from the count, and I am not asserting any of it in either direction.
The point is narrower and harder: a design that provides a contest route and publishes none of its traces has protected the member and left the public exactly where it started.
What the material does not carry
Across the material read on 17 November 2026, no edition of a sanctions transparency report was inspected. The field list, the stated reporting period, the registry's own wording on withholding identities and the presence or absence of any case reference in the reports themselves are all unestablished here. The cadence is not established as quarterly from the registry's own description either, which speaks of defined reporting periods without naming an interval; the only place the word quarterly appears in this material is a third-party characterisation of the latest edition.
And the editions do not reconcile. The entry this briefing addresses is labelled a Q2 2026 report. The only edition described in the material read is one the third-party account calls the Q4 2026 report. Nothing here establishes that these are the same edition, consecutive editions, or editions of the same series, so both labels travel unmerged. What the count adds up to is therefore accurate for the edition described and not transferable to the one named in the label.
What this adds, measured rather than claimed
A search of existing theses for the registry's name at limit 25 on 17 November 2026 returned 25 published rows, every one of them about the registry's general activity — impersonation phishing, governance consultations, meeting registration, training examinations, cooperation agreements, billing schedules, board minutes — and none reads the sanctions transparency report, the sanctions procedure, or any count of frozen resources. The directory entry for the RIPE NCC quarterly sanctions transparency report records no attached article. That is a measurement taken on one day, not a claim of firstness, and none of those 25 rows is a weaker piece of work for not being this one.
The small control that would change everything
Here is what would move this article from its current position. Not names. A registry that publishes who it froze would be committing a different and arguably worse act, and the described privacy and legal ground is a defensible one. What is missing is a line per case that identifies nothing: a decision date, a category of measure, the ground in the sanctions regime's own vocabulary, whether the affected party was notified before the measure took effect, whether a review was requested, and how it ended. Twenty-eight such lines would publish no identity at all and would let a stranger check one freeze against one list entry on one date.
Absent that, the report does exactly half of the work accountability asks for, and it does that half unusually well. It tells you the machinery moved, in quantities precise enough to audit against themselves. It ends the sentence early — at the point where a reader would need the next clause to test whether any particular move was correct. A tally is not a cover-up. It is a count with the checking end cut off, and the cut is where the design says it should be.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






