Stablecoins as Surveillance Objects: the Mirror, Measured
90 OFAC-listed addresses: 90 frozen on Tether, 90 frozen on Circle, 0 of 95 on TRON. We probed both contracts address by address — the mirror, the gap, and what they mean for a no-KYC exit.
Pull the US Treasury's SDN list on the morning of 22 November 2026 and grep it for Ethereum addresses: 41 designated entities carry 90 of them — the treasury wallets of the Lazarus Group, the Garantex Europe OU deposit addresses, and all eight addresses of Roman Semenov, the Tornado Cash developer whose SDN entry still lists [email protected] as his email. Then ask the two largest stablecoin issuers about each address directly, on-chain — no API key, no account, one read-only call per address. Tether's contract answers: frozen on all 90. Circle's contract answers: frozen on all 90. A 100% mirror, executed by two independent contracts run by two competing companies. (Is Tor safe for swaps?)
A mirror, measured: 90 of 90, twice
The method is exactly as boring as the mechanism. We downloaded sdn.csv from treasury.gov, parsed every 'Digital Currency Address - ETH' field into a list of 90 unique addresses, and at 08:01–08:14 UTC on 22 November 2026 called isBlackListed(address) on Tether's USDT (0xdAC17…1ec7) and isBlacklisted(address) on Circle's USDC (0xA0b8…eB48) — the same mapping lookups the transfer functions consult before allowing a send. Ninety returns of true from Tether. Ninety returns of true from Circle. A freshly generated control address read false on both, and the one old Garantex-era address we added as a sanity check — not on today's list — read false too. The maps track the current list, not a decade of grudges. (XMR to USDT exit)
Who is on the list matters as much as how many. It is not one crime: Lazarus Group under the DPRK program, Garantex under Russia-EO14024, the Tornado Cash developer under the same DPRK designation as the 2022 takedown, Iranian individuals, a sanctioned VPN service, a Venezuelan crew — different programs, different prosecutors. What the issuers add is uniformity: they do not curate, they execute. The contract is the last line of a paperwork chain that starts in a designation notice, and on Ethereum that chain now ends with 100% fidelity in both directions — every listed address frozen, and (in our sample) nothing frozen that is not listed. (Is Tor safe for swaps?)
Freeze timing shows the batch habit
Ethereum never timestamps a freeze — the state is a mapping, and reading a mapping leaves no trace. But the chain leaks the moment anyway: a blacklisted address cannot send USDT, so its last successful outgoing transfer brackets the freeze. For Garantex's seven addresses: 0x7FF9…E3D9 last moved 358,003 USDT at 14:55 UTC on 5 April 2022; the remaining six cluster hard — four of them (0x8Dce…, 0xb338…, 0xf437…, 0xD850…) made their final moves within 22 minutes of each other, 22:55 to 23:17 UTC on 25 September 2023, and the last two (0x57EC…, and 0x7CEd…, which moved 1,059,763.84 USDT) the following afternoon at 16:10 and 16:26 UTC. That rhythm is not a court order being served at a desk. It is a spreadsheet being applied. (XMR to USDT exit)
And the switch keeps getting pulled. On TRON's USDT — where the issuer's own events are public — the newest AddedBlackList fired at 22:49:39 UTC on 21 November 2026, block 88,209,084, hours before we ran these probes, with two more additions minutes apart at 22:46 and 22:49. The mechanism is not a dormant emergency power that might someday be used. It is weekly maintenance. (XMR to USDT exit)
The TRON gap
Then run the same probe where most offshore USDT actually moves. Today's SDN file carries 99 TRON-format addresses; 95 of them pass base58check — and exactly zero are frozen on TR7NHq…ed13c as of this morning. The method checks out: the address that was frozen there the day before (TEwhiC82…, block 88,209,084) returns true, so a false is a real false, not a broken probe. Same issuer, same week, opposite answer to the same government list. On the Binance-pegged USDT deployment, the blacklist read functions revert entirely — that switch cannot even be queried under these names. (XMR to USDT exit)
Two readings are possible, and both are worth keeping. The uncharitable one: the TRON deployment — the rail for most high-volume USDT — lags the sanctions list, and a designated entity could still move TRC20 USDT as of this morning. The honest one: mirroring is a per-deployment human decision, not protocol sync, and a gap means 'not yet', never 'clean'. There is a third finding hiding in the data quality: Garantex's TRON entry in the SDN file is a 26-character string that fails base58check, two more entries carry trailing periods, and one is 20 characters. The government's own file is dirty, and whoever applies it by hand has to guess — which is very likely how a gap between two contracts of the same issuer begins. (XMR to USDT exit)
What the mechanism means for the exit path
For a no-KYC swap the exposure is directional, and it is worth saying plainly: issuer coins are the only assets in the route that a third party can rewrite to zero after you already hold them. A mapping flips, every transfer from the address reverts, and destruction is the documented follow-up — we measured 200 sampled burns totalling 93,419,218.93 USDT in October. Monero has no owner key and no blacklist mapping; Bitcoin has neither. So a stablecoin leg is fine as a bridge in motion — and a bad place to stop. (XMR to USDT exit)
Our own prices that same morning, measured 08:14–08:24 UTC against Kraken's XMR/USD mid of 481.17: 1,000 USDT-TRC20 bought 2.03321 XMR (an implied premium of +2.21%), 5,000 bought 10.172884 (+2.15%), and 100,000 USDC-ETH bought 202.91456263 (+2.42%, a single locked route with a 120-second window). The exit side was flat: 1 XMR fetched 481.298 USDT-TRC20, +0.03% against mid. The round trip cost 2.14%. That premium is the fee for not handing over your identity; the freeze function is the reason the last hop of an exit should not end on an issuer's token. Enter, swap, leave. (XMR to USDT exit)
Two cheap habits follow from the data. Before any operation with a stablecoin you received elsewhere, read the mapping for your own address — it is one read-only call and the same check the transfer function will make when you try to move it. And calibrate expectations for relief: removals do happen on the issuer's schedule (447 addresses released inside a single hour on 14 May 2026), but in the 200 destructions we sampled, the funds did not come back. (XMR to USDT exit)
What this is not
Ninety of ninety is the designated universe, not a draw from ordinary users — every normal address we probed read false, and that is the norm. Keep the two layers apart: an issuer freeze is binary, on-chain, and takes effect at the contract; exchange risk scoring is probabilistic, off-chain, and takes effect at a deposit desk — most holders who get stuck will meet the second layer, not the first. What today's test adds is that on Ethereum the binary layer now tracks the government list with no exceptions in either direction. This is measurement, not legal advice. Designations are decisions made elsewhere and rewritten into code that cannot argue.
Closing ledger. 22 November 2026: 90 OFAC-listed Ethereum addresses, 90 frozen on Tether, 90 frozen on Circle, 0 of 95 TRON-listed addresses frozen on TRON — all measured in one morning with read-only calls anyone can repeat. Entering XMR that same morning cost +2.15% to +2.42%; leaving cost +0.03%; the round trip 2.14%. Stablecoins became surveillance objects not because watching got better, but because the endpoint moved: the output of a designation used to be a report, and now it is a mapping you can query but cannot edit. Quotes move — check the live page. The mapping does not unwrite itself.