The Other Half of the Grid
The American surveillance grid left a question open and named it: what the partners are permitted to do. Here are nine British powers under the same four questions, and the row with no instrument behind it at all — material a partner collected and passed on. The laundering claim turns out to be half answered by a statute almost nobody quotes, and half not answered at all.
The question the American grid refused
The surveillance grid set eight American authorities against four questions and declined to open a fifth: what the partners are permitted to do. It named that as a subject of its own. This is it.
The arrangement has a name and a date. The UKUSA agreement was signed in March 1946, grew out of a 1943 wartime accord, and was extended to Canada in 1948 and to Australia and New Zealand in 1956. It was not acknowledged in public for most of its life. The text was declassified by the NSA and GCHQ together on 24 June 2010, covering 1940 to 1956, which is the unusual part: the founding document of the thing most often described as unaccountable can be read online, and the agencies published it themselves.
What cannot be read is anything equivalent about the present. So the grid above does what the American one did — takes the powers one at a time and asks the four questions of each — and then adds the row that has no instrument behind it.
The British scheme constrains reading, not taking
The single structural difference from the American arrangement is worth stating before anything else, because most arguments about British surveillance are conducted without it.
American law mostly draws its line at collection: Section 702 may not target a US person, and the protections attach to who may be acquired. The Investigatory Powers Act draws its line in a different place. Bulk interception is permitted where the main purpose is to obtain overseas-related communications — and section 136 defines that as communications sent or received by individuals outside the British Islands, so one party abroad is enough. A British resident's messages are inside the definition whenever the person at the other end is not here.
What then protects the resident is not a limit on what was collected but a prohibition on reading it. Under section 152, content taken in bulk may not be selected for examination using criteria referable to an individual known to be in the British Islands where the purpose is to find out what that person sent or received. Lifting that prohibition requires a targeted examination warrant, which goes through exactly the same double lock as a warrant to intercept the person directly: a Secretary of State, and a Judicial Commissioner applying judicial review principles.
There is one exception, and it has a clock on it. If the person enters the British Islands, a senior officer may authorise examination in writing without a warrant, but only until the end of the fifth working day after the change of circumstances, and the Secretary of State must be told.
Whether reading-side control is weaker or stronger than collection-side control is a real argument. It is not the argument people usually have, which tends to assume the British scheme has no equivalent of the American rule at all.
The laundering claim, which is half answered
The claim that gives this subject its energy is specific: that partners spy on each other's nationals and trade the results, so that each service obtains by alliance what it is forbidden to collect at home.
Half of that is addressed by a provision almost nobody quotes. Section 9 of the Act is headed Restriction on requesting interception by overseas authorities, and it says that a person in the United Kingdom may not ask an overseas authority to intercept the communications of someone they believe will be in the British Islands unless a targeted interception or targeted examination warrant already authorises that interception here. The request route is closed by the same instrument that would have been needed to do it directly. The interception code of practice goes further and requires the UK authority to hold a warrant where the overseas authority would not otherwise have been intercepting that person at all, with the material then handled under the same internal safeguards as material collected under the Act.
The other half is not addressed, and the grid's last row is where it sits. There is no warrant to receive intelligence. If a partner collected material under its own law, for its own reasons, and a British resident is in it, no British instrument governs the handover. Section 9 regulates asking. Nothing regulates arriving.
What the courts have said about the half that is open
This is not an untested gap. It has been litigated twice, and the two results point in different directions, which is the honest shape of it.
In February 2015 the Investigatory Powers Tribunal held that the regime governing British access to material from the American PRISM and Upstream programmes had been unlawful until December 2014 — not because the sharing was improper in itself, but because the rules governing it were secret, and a secret rule cannot satisfy the requirement that an interference with private life be in accordance with law. It became lawful at the moment the policies were disclosed during the proceedings. It was the first time in the Tribunal's fifteen-year history that it ruled against the intelligence services.
In May 2021 the Grand Chamber of the European Court of Human Rights decided the other way on the same question. It found the old RIPA bulk interception regime in violation of Articles 8 and 10 — authorised by a Secretary of State rather than a body independent of the executive, with search terms linked to an individual not subject to prior independent authorisation — while holding, by twelve votes to five, that the regime for requesting and receiving intelligence from foreign partners did not violate Article 8.
So: the sharing arrangement was unlawful while its rules were secret, and lawful once they were published, and a court of seventeen judges split five ways on whether the published version is enough. Anyone who tells you this question is settled in either direction is describing one of those rulings and not the other.
The column that does not vary
The American grid found that one authority out of eight notified its targets, and that this single provision explained the shape of everything beneath it.
The British answer is simpler. There is no notification requirement in the Act at all. Nine powers, one answer, nine times. The only route by which a person ever learns they were intercepted, interfered with or searched is section 231: the Investigatory Powers Commissioner may inform them of a relevant error, but only if the Commissioner considers it serious, and an error is not serious unless it caused significant prejudice or harm — the Act says in terms that a breach of the person's Convention rights is not enough on its own. Even then, the person is told only if that is in the public interest, and the Commissioner may not tell them by any other route.
The practical consequence is the one this archive keeps running into. A British resident who suspects they were surveilled cannot find out. They can complain to the Tribunal, which investigates in private and, where it finds nothing improper, returns a determination that does not disclose whether there was any surveillance at all.
What people do when nobody is ever told
If the state will not tell you, the remaining route is records law — and the clearest demonstration of what that produces is not British at all. It is American, it concerns a private company rather than an intelligence agency, and it is worth putting next to section 231 because the contrast is so stark.
Flock Safety sells automated number-plate recognition to American police forces and private communities: cameras on public roads, images to the company's servers, plates and vehicle characteristics indexed into a database shared across its customers. When an officer runs a plate, the system records who searched, when, and the reason they typed in. Those audit logs are government records, which in many American jurisdictions makes them obtainable.
Three independent projects have been built on that opening, and none of them is affiliated with the company.
- Have I Been Flocked aggregates audit logs released through open-records requests and transparency portals, and lets anyone check whether their own plate appears in them. Its published figures as of early October 2026 are 243,098,172 searches covering 4,686,527 distinct plates, drawn from 6,598 organisations, with the earliest record from December 2021. The project says plainly that the dataset is incomplete and lagging — few agencies release the logs readily, what arrives is often redacted, and a result shows that somebody queried your plate, not that you were stopped or investigated. It also states the fact that makes the rest of it matter: many agencies require neither a warrant nor suspicion before running a search.
- DeFlock maps where the cameras are. It is a volunteer project begun in late 2024 that writes its pins into OpenStreetMap rather than a private database, so the map it produces belongs to nobody. Tallies of ALPR-tagged features in that dataset ran to roughly 128,000 in the United States by August 2026, against the company's own estimate of 120,000-plus devices.
- Eyes on Flock works the other seam, aggregating and analysing the public transparency portals the company provides to its customers, to track cameras, searches and data sharing between agencies.
All three are advocacy projects and say so, and their numbers are floors rather than counts: they describe the portion of a private system that public-records law has pried loose, which is why the totals move whenever another force answers a request.
Set that against the row above. A British resident subject to any of the nine powers on this grid learns nothing, ever, unless the Commissioner decides an error was serious enough. An American motorist can type a plate into a website and find out how many times it was queried, by which agency, and what reason the operator typed — for the subset of forces whose records someone has already extracted.
Neither of those is accountability in the sense the word usually implies. One is silence by statute. The other is a volunteer reconstruction of a commercial database, assembled a records request at a time. But only one of them produces a number, and the gap between them is the practical meaning of a notification regime that exists only for errors.
What this grid cannot establish
It describes authorities, not conduct. Every row says what the Act permits and who must approve it; none of them says what was done last year, because that is not public and the grid would be inventing it.
It covers one partner of five. The other four — the American authorities are in the companion grid, and Canada, Australia and New Zealand have statutes of their own — are not here, and the arrangement between them is governed by instruments that have not been published for the period that matters.
And it cannot tell you the size of anything. The constraint that bites in this scheme is at the point of examination, which is exactly the point at which no public figure exists: the number of targeted examination warrants is not broken out in a way that answers the question people are actually asking, and the volume of material arriving from partners is not reported at all.
What the record does establish is narrower and worth holding onto. The request route is closed by statute. The receipt route is not closed by anything. One tribunal found the receipt regime unlawful for as long as its rules were secret, and one court found the published version compatible by a majority of twelve to five. And whichever way that question goes, nobody affected by it will be told.
Sources
- Investigatory Powers Act 2016, section 9 — Restriction on requesting interception by overseas authorities
- Investigatory Powers Act 2016, section 136 — bulk interception warrants, and the definition of overseas-related communications
- Investigatory Powers Act 2016, section 152 — the prohibition on selecting for examination the content of a person known to be in the British Islands, and the five-working-day exception
- Investigatory Powers Act 2016, section 199 — the statutory definition of a bulk personal dataset, including that the majority of individuals in it are not and are unlikely to become of interest
- Investigatory Powers Act 2016, section 231 — error reporting, the serious-error threshold, and the bar on informing a person except as that section provides
- The Powers — the Investigatory Powers Commissioner's own account of which powers are double locked and who authorises each
- Interception of communications: code of practice — the requirement that a UK warrant be in place before asking an overseas authority to intercept a person here, and the notification of unwarranted requests to the Commissioner
- Big Brother Watch and Others v the United Kingdom — Grand Chamber judgment of 25 May 2021: the RIPA bulk interception regime in violation of Articles 8 and 10, and the foreign-intelligence-sharing regime held compatible by twelve votes to five
- Investigatory Powers Tribunal ruling of 6 February 2015 — British access to PRISM and Upstream material unlawful until December 2014 because the governing rules were secret; the first ruling against the agencies in the Tribunal's history
- Declassified UKUSA signals intelligence agreement documents, released jointly by NSA and GCHQ on 24 June 2010 and covering 1940–1956
- A brief history of UKUSA — GCHQ's own account of the 1946 agreement and its extension to Canada, Australia and New Zealand
- Have I Been Flocked — aggregated Flock Safety audit logs obtained through open-records requests and transparency portals, reporting 243,098,172 searches across 4,686,527 distinct plates as of 2 October 2026; fiscally sponsored by the Alternative Newsweekly Foundation and unaffiliated with the company
- DeFlock — a volunteer project mapping automated number-plate readers into OpenStreetMap rather than a private database, begun in late 2024
- Eyes on Flock — an independent project aggregating and analysing data from Flock Safety's public transparency portals, tracking cameras, searches and data sharing across the United States