Public-Interest Investigation · The Capstone · June 2026
The Whole Picture: What It All Adds Up To
The capstone — one place, everything we found and how it actually connects. Two machines were documented across this work: one that surveils and sorts populations, one that persuades them. They are not a single conspiracy — and I won’t claim they are. What binds them is a pattern, and the pattern is why it’s not good: the systems that surveil, sort and persuade whole populations increasingly run with the oversight switched off. Built from public records, tiered, and as careful about what is not connected as about what is.
The whole picture — two machines, one pattern
Everything documented, in one frame. Two different machines — not a single conspiracy. What binds them is a pattern, and the pattern is the story.
Oversight off — sole-source, secret, sub-threshold; no impact assessments
Foreign at the ends — foreign-owned operational layer; foreign-funded persuasion
AI is the frontier — the least-visible layer where both sorting & shaping now happen
No one was asked — the prior question never put to the people it's done to
What is NOT connected — stated as plainly as the rest
• No Parscale–Palantir tie. Tested directly — clean negative.
• No Thiel / Silicon-Valley cabal. Run against ~85 names: one documented tie (Kushner), the rest empty. No hidden web.
• Allegations stay allegations (Gaza, the “mega-database,” AI effectiveness); myths corrected (not Stop the Steal; the CA legend; WinRed/Brazil were others).
The capstone in one image: two documented machines (surveil/sort and persuade), the structural pattern that binds them (oversight off, foreign at the operational and persuasion ends, AI the new frontier), and — the part that makes it hold — what was tested and found NOT connected.
Two different machines have been documented here. One sorts and surveils populations (record-linkage and the operational data platforms beneath government). The other persuades them (foreign-funded influence operations, now reaching into AI itself). They are not a single conspiracy, and I will not claim they are — that overstatement is the line that would discredit everything true beside it. What they share is a pattern, and the pattern is the story:
The most consequential systems — the ones that surveil, sort, and persuade whole populations — increasingly run with the oversight switched off. Under-threshold or sole-source contracts. No published impact assessments. Foreign ownership or foreign funding at the operational and persuasion ends. Disclosure, where it exists at all, that almost no one is told to look for. And the prior question — should this happen, and were the people it is done to ever asked? — was never put.
That is why it's not good. Not because of a hidden hand, but because of an accountability vacuum that keeps showing up in the same shape, in system after system.
Half one — the data state (surveillance & record-linkage)
How the modern state joins up the separate records it holds on you, and who runs the operational layer.
The Splink Investigation — the UK Ministry of Justice's open-source record-linkage tool, stitching health, tax, justice, benefits and census records into person-level profiles, lawfully, with no consent, no notification, no opt-out — and quietly sliding from statistics into real-time operations (a daily probation→police feed). No court ruled it unlawful; the same policy-not-law shortcut was struck down three times next door (the DPA 2018 immigration exemption).
Palantir: The Global Footprint — the US national-security firm that is the operational data/AI layer of the state's hardest functions worldwide (targeting, deportation, health, policing), procured repeatedly sole-source or secret. A constitutional court (Germany, 2023) has already ruled the underlying powers unlawful; €850m+ has been divested; and the allegations it denies (Gaza, a US "mega-database") are kept labelled, not asserted.
The US Database — how ICE joins up government, biometric and commercial records to locate people for removal: the Palantir-built case engine, the data-broker contracts, the $4.3bn biometric backbone, and the watchdog finding ICE refused to concur with.
The City of London — a public authority unaccountable by design: a private fortune beyond Freedom of Information, a business franchise abolished everywhere else, and the national fraud database it runs on a platform the public can't fully see.
Half two — the influence machine
How opinion itself is now sorted and shaped — increasingly by foreign money, and increasingly aimed at AI.
The Parscale Investigation — Trump's former campaign manager, now a foreign-government influence contractor. The honest frame, held throughout: he builds a deceptive persuasion machine, for hire, foreign and domestic — not the disinformation mastermind of legend. His own documented deception sits in the 2020 ads he ran; the rest of the legend (Stop the Steal, the Cambridge Analytica myth) is corrected, not repeated.
Foreign Agent for Israel — his FARA-registered Clock Tower X: a disclosed Israeli-government AI-influence apparatus (via Havas), built in its own filing to bend ChatGPT and Claude. Our own infrastructure check proved its "independent" sites are one operator — labeled, not hidden; the story is coordination, self-dealing, and intent, not concealment.
The Machine Goes South — the same data-targeting machine exported into elections in Argentina, Bolivia and Honduras, in partnership with a documented disinformation operator (Fernando Cerimedo). Parscale built the machine; the lies are Cerimedo's record — and in no country is the disinformation forensically tied to Parscale.
The First Wave — the bigger frame: states have begun buying the manipulation of AI outputs. Israel/Parscale is the first FARA-disclosed case; Russia's "Pravda" network the only state peer; a commercial "generative engine optimization" industry and a settled poisoning science enable it — while the AI platforms stay publicly silent. Whether it works is genuinely contested, and that caveat is kept in full view.
What actually connects them (and what does not)
The honest connective tissue is thematic and structural, not a secret network:
The oversight is off in the same ways. Sole-source/secret procurement and missing impact assessments run through both halves — the redacted NHS-Palantir contract and the £489,999 Met pilot on one side; the disclosed-but-unscrutinised FARA influence contracts on the other.
A foreign operational/persuasion layer. The pattern repeats: the high-risk, identifiable, operational end — and now the persuasion end — is foreign-owned or foreign-funded (a US firm holding sovereign data; a foreign government paying US operatives to shape opinion and AI).
AI is the new shared frontier. Palantir's operational AI sits at the action end of the data state; the First Wave sits at the persuasion end. The same technology is becoming the layer where both sorting and shaping happen — with the least public visibility of all.
The same discipline binds the reporting. Every part was built to tier evidence and refuse the unsupported arrow. That is the connective tissue that makes the whole hold.
What is NOT connected — stated as plainly as the findings, because this is what makes it trustworthy:
Parscale is not connected to Palantir. Tested directly — no contract, data, personnel or money tie. A clean negative.
There is no Parscale–Thiel / Silicon-Valley-cabal web. Run against ~85 names from a circulated "elite list": exactly one documented business tie (Jared Kushner, already known); the rest empty. There is no hidden network here.
The most sensational allegations are held as allegations. Gaza targeting, a US "mega-database," AI-manipulation effectiveness — labelled, denied where denied, and never asserted as proven.
The myths are corrected, not amplified. Parscale did not run Stop the Steal; the Cambridge Analytica "psychographics won 2016" story is overstated; the WinRed dark patterns and the Brazil election hoax were other people's work.
Why it's not good — the verdict
Read the two halves together and the sober conclusion isn't a cabal — it's worse in a quieter way: the machinery to surveil, sort, and now persuade entire populations has been built and deployed, lawfully or in disclosed-but-unscrutinised form, with the public asked about none of it — and the foreign-owned and AI-mediated ends of it are the least visible and least governed of all. No single crime explains it. The accountability gap does. A state — and a politics — that can join up and act on three in four of its people, and pay to shape what their AI tells them, does not need to break the law to be worth worrying about. It only needs us not to look.
What to do about it
This was never an argument that the maths is wrong or that someone broke the law. It is an argument that the prior question was never asked. So put it on the record: ask your representative which of your records have been linked and whether you can opt out; FOI the unredacted contracts and the undisclosed FARA supplements (the Bridges influencer roster is one missing document away); ask the platforms, on the record, what they are doing about AI-output manipulation; and press for the one reform that fits every case here — that population-scale linkage, operational integration, and foreign-funded influence require an individual right to know, and to object, not merely a lawful basis or a buried disclosure.
Method: built from public sources only; every claim tiered and sourced in the underlying reports; allegations labelled and not asserted; living individuals treated with the documented-not-insinuated standard; the stories others broke credited to them. Full evidence, primary filings, graphics and archived sources: the public repository at github.com/btmaffiliate/uk-data-linkage-disclosure. Not legal advice.