Dossier mode
The Cost/Benefit of Financial Surveillance
The same investigation, restaged one beat at a time. Drive it with the arrow keys, space, or autoplay. Nothing is cut from the piece — long runs are split across frames. Read the full investigation or open the Surveillance State(s) hub.
The Cost/Benefit of Financial Surveillance
A machine that watches essentially every transaction on earth, costs $200B+ a year, and — by its own metric — intercepts a fraction of a percent of criminal money.
To stop dirty money we built one of the largest surveillance systems ever, and by its own confiscation metric it catches almost none of it.
This is a cost-benefit question, not an accusation. Every figure is attributed to a named study. The cost and the seizure rate are FACT. The claim that they prove the system 'fails' is contested — and we grade it that way.
Share of criminal proceeds the anti-money-laundering system seizes, by its own confiscation metric — ~0.1% (Pol 2020), ~0.2% (UNODC 2011), ~1% confiscated in the EU (Europol 2016).
Pol 2020 · UNODC 2011 · Europol 2016
The documented spine.
- Under the Bank Secrecy Act and its global counterparts, banks must file Suspicious Activity Reports and currency-transaction reports, screen against sanctions lists, and monitor customer activity continuously. FinCEN receives millions of SARs and CTRs a year; the FATF exports the model to virtually every country.
- It costs more than $200 billion a year to run. LexisNexis Risk Solutions' 'True Cost of Financial Crime Compliance' study put global spending around $206 billion in 2023, borne overwhelmingly by financial institutions and, through them, their customers.
- It observes ordinary financial life down to the individual transaction. Whatever else it is, it is one of the largest routine surveillance systems ever built.
By its own confiscation metric, the system intercepts on the order of 0.1% to 1% of criminal proceeds.
The estimates are consistently tiny, though they vary with what is counted: Ronald Pol's 2020 paper puts recovery near 0.1%, the UN's 2011 study near 0.2%, and Europol's 2016 review about 1% confiscated in the EU. We report the range and attribute each estimate — we do not launder Pol's 0.1% into 'the' number.
“The anti-money laundering system may be the world's least effective policy experiment.”
Policy Design and Practice, 2020 — the much-cited paper behind the ~0.1% recovery figure.
Everyone's transactions are watched, the poor get de-banked, and the institutions that moved cartel and oligarch billions pay a fine.
Two documented downstream effects fall on the powerless: 'de-risking,' in which banks drop whole categories of customers — money-service businesses, charities, remittance corridors, entire regions — rather than bear the compliance risk (a concern the World Bank, IMF, and FATF have all acknowledged); and civil asset forfeiture, the AML-adjacent power to seize money on a reduced burden of proof. Meanwhile the largest institutional launderers, when caught, resolve with fines and deferred prosecutions.
The system's data-center environmental footprint is real, riding a steep growth curve — and not separately measured.
Transaction monitoring and sanctions screening run continuously across vast data-center infrastructure. The broader sector's footprint is measured — the IEA puts data centers at ~415 TWh in 2024 (about 1.5% of world electricity), roughly doubling by 2030, and LBNL estimated tens of billions of litres of US cooling water in 2023. What no public accounting isolates is the slice attributable to financial surveillance specifically. So we do NOT assert the viral '3 TWh / 15 billion litres' figures — we could not source them; the provable claim is that the cost is real and unmeasured.
Defenders argue the value is deterrence and intelligence, not seizure rate — and a low confiscation percentage cannot capture that.
This is the strongest counter-argument, and it is a real one. Regulators and many AML professionals argue the regime's value lies in raising the cost and friction of laundering, in the financial intelligence SARs feed to law enforcement, and in prevention — none of which shows up in a 'money recovered' statistic. On this view a low confiscation percentage is a category error, not an indictment. The confiscation numbers are documented FACT; the claim that they prove the system 'fails' is contested, and the deterrence value is genuinely hard to measure either way.
What the record establishes, and what it does not.
- The cost (>$200B/yr) and the low confiscation rate (~0.1-1%) are sourced, attributed facts.
- De-risking and civil forfeiture fall hardest on the poor and the financially excluded.
- The biggest institutional launderers, when caught, settle for fines.
- That the system is a scam, a conspiracy, or deliberately designed to fail. Expensive and low-yield is not the same as sinister.
- That a low seizure rate, by itself, proves the system is worthless — its defenders' deterrence argument is real and hard to measure.
- The specific viral energy and water figures (3 TWh / 15B litres). We could not source them, so we do not assert them.
The share of the data-center energy and water footprint attributable to financial surveillance specifically is unmeasured.
A public accounting that isolates the financial-surveillance slice of the IEA's ~415 TWh and LBNL's cooling-water figures, rather than the sector aggregate.
Help us fill it →Why it matters now.
This is where the Surveillance State(s) hub meets Too Big to Jail. A system justified as crime-fighting subjects everyone's financial life to continuous monitoring, at a cost of hundreds of billions, and reaches a fraction of a percent of criminal money — while the de-banked poor bear the cost and the banks documented to have laundered for cartels and oligarchs, HSBC and Deutsche among them, settle for fines. Whether the regime deters what it cannot seize is a real, open question. Whether it is aimed at the right targets is the one this hub keeps asking.
Help us fill these lines.
- OpenWhether the AML regime's deterrence and intelligence value — which its own defenders concede a confiscation percentage cannot measure — actually justifies $200B+ a year, when the seizure rate its own metric reports runs just 0.1% to 1%.
- OpenWhat share of the data-center energy and water footprint (the IEA's ~415 TWh, LBNL's cooling-water figures) is attributable to financial surveillance specifically — a slice no public accounting isolates.