← Verdix

Base address poisoning — October 2026

A monthly report from Verdix's poisoning watcher, which reads every token transfer on Base as blocks arrive. Data from Sep 23 (when the watcher started) through Oct 19, 2026, end of day UTC; the month isn't over yet, so October figures are month to date.

1,078,205
new lookalikes, 1–19 October (month to date)
56,748 per day on average
1,820,544
flagged in total
since Sep 23, 2026
7,545,783
poisoning transfers seen
on flagged lookalikes, all time
926
fake token contracts flagged
770 of them new in October
3 · $2,080
real payments to lookalikes we had already flagged
since the watcher started

New lookalikes per day

Fake-token baitZero-value bait
New address-poisoning lookalikes flagged per UTC day on Base Stacked bars: fake-token bait and zero-value bait. The table below has every value. 0 50,000 100,000 150,000 200,000 2026-09-23: 1,976 new (0 fake-token, 1,976 zero-value) 2026-09-24: 59,605 new (16,327 fake-token, 43,278 zero-value) 2026-09-25: 127,902 new (119,243 fake-token, 8,659 zero-value) 2026-09-26: 63,325 new (53,798 fake-token, 9,527 zero-value) 2026-09-27: 145,119 new (121,741 fake-token, 23,378 zero-value) 2026-09-28: 63,458 new (32,436 fake-token, 31,022 zero-value) 2026-09-29: 172,506 new (149,772 fake-token, 22,734 zero-value) 2026-09-30: 108,448 new (91,839 fake-token, 16,609 zero-value) 2026-10-01: 72,404 new (28,929 fake-token, 43,475 zero-value) 2026-10-02: 42,552 new (27,575 fake-token, 14,977 zero-value) 2026-10-03: 39,050 new (24,019 fake-token, 15,031 zero-value) 2026-10-04: 37,657 new (26,572 fake-token, 11,085 zero-value) 2026-10-05: 91,727 new (64,855 fake-token, 26,872 zero-value) 2026-10-06: 60,480 new (35,361 fake-token, 25,119 zero-value) 2026-10-07: 57,922 new (39,515 fake-token, 18,407 zero-value) 2026-10-08: 62,524 new (61,531 fake-token, 993 zero-value) 2026-10-09: 78,422 new (30,147 fake-token, 48,275 zero-value) 2026-10-10: 75,519 new (47,205 fake-token, 28,314 zero-value) 2026-10-11: 62,041 new (53,448 fake-token, 8,593 zero-value) 2026-10-12: 55,917 new (55,141 fake-token, 776 zero-value) 2026-10-13: 44,363 new (43,630 fake-token, 733 zero-value) 2026-10-14: 34,605 new (33,412 fake-token, 1,193 zero-value) 2026-10-15: 22,680 new (21,883 fake-token, 797 zero-value) 2026-10-16: 71,167 new (70,665 fake-token, 502 zero-value) 2026-10-17: 56,896 new (55,730 fake-token, 1,166 zero-value) 2026-10-18: 43,054 new (41,217 fake-token, 1,837 zero-value) 2026-10-19: 69,225 new (68,677 fake-token, 548 zero-value) Sep 23 Oct 06 Oct 19

UTC days, each address counted once, on the day we first flagged it. The dashed line marks October 1. In October so far the busiest day was Oct 05 (91,727) and the quietest Oct 15 (22,680).

Table: every day
Day (UTC)New lookalikesFake-token baitZero-value bait
2026-09-231,97601,976
2026-09-2459,60516,32743,278
2026-09-25127,902119,2438,659
2026-09-2663,32553,7989,527
2026-09-27145,119121,74123,378
2026-09-2863,45832,43631,022
2026-09-29172,506149,77222,734
2026-09-30108,44891,83916,609
2026-10-0172,40428,92943,475
2026-10-0242,55227,57514,977
2026-10-0339,05024,01915,031
2026-10-0437,65726,57211,085
2026-10-0591,72764,85526,872
2026-10-0660,48035,36125,119
2026-10-0757,92239,51518,407
2026-10-0862,52461,531993
2026-10-0978,42230,14748,275
2026-10-1075,51947,20528,314
2026-10-1162,04153,4488,593
2026-10-1255,91755,141776
2026-10-1344,36343,630733
2026-10-1434,60533,4121,193
2026-10-1522,68021,883797
2026-10-1671,16770,665502
2026-10-1756,89655,7301,166
2026-10-1843,05441,2171,837
2026-10-1969,22568,677548

By week

Week (Mon–Sun)New lookalikesPer dayFake-tokenZero-value
Sep 23–Sep 27 (partial)397,92779,585311,10986,818
Sep 28–Oct 04536,07576,582381,142154,933
Oct 05–Oct 11488,63569,805332,062156,573
Oct 12–Oct 18328,68246,955321,6787,004
Oct 19–Oct 19 (partial)69,22569,22568,677548

What the bait looks like

A poisoner needs the lookalike address to show up in your wallet history, next to the real one. On Base we see two ways of doing it:

A third kind, dust (a tiny real amount sent from the lookalike), is also used, but we don't flag on it alone, so it isn't counted here. One bot transaction often carries several kinds at once; each address is counted under the kind we saw first. Since mid-October almost all new lookalikes come with fake-token bait; zero-value bait is still sent, but mostly to lookalikes that are already flagged.

Why the first and last four characters aren't enough

Wallets usually shorten an address to its start and end, like 0x1a2b…9f0e. Poisoners generate a fresh address that matches exactly those characters: the first 4 and last 4 hex characters after 0x, 8 characters in all. That is about 4.3 billion possibilities, which ordinary GPU software works through quickly. The middle 32 characters are different, and that's the part people don't read.

Our watcher keeps, for every wallet, the counterparties it really exchanged $1 or more of USDC or USDT with in the last 7 days. When a bait transfer links that wallet to an address that matches one of those counterparties 4+4 but isn't it, the address is flagged, in the same block.

Real money, already flagged

Since the watcher started, we found 3 payments of real stablecoins, $2,080 in total, sent to lookalike addresses that were already in our feed, between 32 hours and 6.5 days before the money moved. A wallet or exchange that checked the recipient against the feed could have warned the sender before they signed.

Paid (UTC)AmountFlagged before the paymentChecked on-chain by hand
Oct 02$101.9032 hoursyes
Oct 09$20.154.5 daysyes
Oct 16$1,958.206.5 daysyes

Amounts are what was sent. Sender and lookalike addresses are left out on purpose.

The largest case: on October 9 a Base user paid a regular counterparty. Within minutes a bot planted fake copies of that payment in the user's history, sent to addresses sharing the counterparty's first and last four characters, and refreshed them every day. We flagged the lookalike 29 minutes after the real payment. On October 16 the user copied the wrong address from their history and sent it 1,958 USDC: 6.5 days after our flag. In another case a lookalike we had flagged because it was aimed at one person was paid by a different person who used the same real counterparty.

No wallet used our data at the time, so these are losses that could have been prevented, not ones that were. Three cases are examples, not a statistic.

How to protect yourself

  1. Never copy an address from your transaction history. Take it from the recipient directly, or from your wallet's saved address book.
  2. Compare the whole address, not just the first and last few characters. Poisoners match those on purpose.
  3. After a test payment, don't re-copy the address. Poisoners strike between the test and the real payment, often within minutes. Reuse the address you already verified.

Check a Base address for free →

For businesses: real-time feed / API of flagged lookalike addresses, for wallets, exchanges and payout teams → lauras@verdixapi.com

Method and limits