Address Poisoning: The Lookalike Address Scam Hitting Wallet History
The scam that needs you to do one lazy thing
Address poisoning doesn’t touch your keys. It doesn’t ask you to sign anything strange, connect to a fake site or approve a contract. It waits for you to do something you already do dozens of times a week: open your wallet’s activity list, find the address you sent funds to last time, and copy it.
That’s the whole attack. Somebody plants an address in your history that looks like one you use, and hopes that the next time you’re in a hurry you grab theirs instead of yours.
It works on careful people. That’s the uncomfortable part. You can have separate seeds, a hardware wallet for the real money and a clean approval list, and still send a claim’s worth of tokens to a stranger because two strings of hex looked the same at a glance. Nothing on-chain can reverse that transfer. There’s no dispute button.
Farmers are a better target than most people, for a simple reason. A normal holder might send funds out a few times a month to a handful of places. Someone running a multi-wallet farm is moving gas between wallets, withdrawing from exchanges, sweeping small balances and depositing claimed tokens somewhere to sell, often across several chains in one sitting. That’s a lot of copying and pasting, and every paste is a chance to pick the wrong line.
This piece walks through how the lookalike gets made, how it lands in your history without your involvement, why wallet interfaces make it easy to miss, and the handful of habits that close the gap. The habits are dull. They also work, which is more than can be said for most security advice.
What the attacker is actually counting on
Every Ethereum-style address is forty hexadecimal characters after the 0x. Nobody reads forty characters. Wallets know this, so almost every interface shortens addresses to something like the first four or six characters, an ellipsis, and the last four. Explorers do a version of the same thing in their transaction lists. Your eye learns to check those two ends and trust the middle.
Address poisoning is built on exactly that shortcut. The attacker doesn’t need your address. They need the address of someone you send to, and they need a new address whose first few and last few characters match it. If you normally send to an address that starts 0x7a3F and ends 91cE, they generate one that also starts 0x7a3F and ends 91cE, with a completely different middle.
Then they need that fake to appear somewhere you’ll copy from. Your transaction history is the obvious place, because it’s where people go to find “the address I used last time”.
So the scam relies on three behaviours at once, and it only needs all three to line up once:
- You send to the same counterparties repeatedly.
- You find those counterparties by scrolling history rather than from a saved source.
- You check an address by looking at its ends.
If any one of those is broken, the attack mostly fails. That’s useful, because it means you don’t need perfect vigilance. You need to break one link reliably, and preferably two.
How the lookalike gets generated
Making an address that matches a chosen prefix and suffix is a brute-force search. An address is derived from a private key, and there’s no way to work backwards from the address you want to a key that produces it. So a generator just creates keys very quickly, checks each resulting address against the pattern, and keeps going until one fits.
The cost of that search grows fast with every extra character. Each hex character has sixteen possible values, so matching one more character multiplies the expected work by sixteen. Matching four characters at each end is cheap on ordinary hardware. Matching many more gets expensive quickly. That’s why lookalikes in the wild usually match the parts an interface shows you and not much else: they’re matched to what you’ll see, not to the whole address.
This has a practical consequence. If your wallet shows six characters at each end and you check all of them, a lazy attacker’s four-and-four lookalike fails that check. A more patient one might match more. There’s always a number of characters where the attacker’s cost becomes silly, and it’s well short of forty. Checking a few characters from the middle of an address as well as the ends pushes you past what these generators are built to fake.
There’s also a timing element. Attackers generally run this against many victims at once, watching the chain for transfers and firing a matching lookalike at whoever just sent something. Your address isn’t special to them. You’re one of many targets drawn from recent activity, and the generation work gets spread across all of them.
How it lands in your history without you doing anything
The part that surprises people is that the poisoned entry can appear in your history as if you sent it. There are a few common variants, and it helps to know all of them because they look different in a wallet.
The first is a dust transfer. The lookalike address sends you a tiny amount of a real asset, so your history now shows an incoming transfer from an address that resembles someone you deal with. It costs the attacker a little gas and a little dust per target.
The second is the zero-value transfer. Standard fungible token contracts have a function that lets an approved spender move tokens from one address to another. Many implementations let anyone call it for an amount of zero without any approval, because moving nothing from your balance was never considered a risk. The result is a transfer event that says tokens moved from your address to the lookalike, for zero. Your wallet and the explorer read the event and list it in your activity as an outgoing transfer. You didn’t sign it. It still sits in your history, pointing at the attacker’s address, looking like something you did.
The third is the counterfeit token. The attacker deploys their own token with the same name and ticker as a well-known stablecoin, then sends a matching amount of it from the lookalike. If you recently sent 250 of a stablecoin to a friend, you may see what looks like a similar transfer of the same ticker, but from a worthless contract.
None of these can take anything from you directly. They’re bait placed in the one list you trust most.
The zero-value variant deserves a second look because it’s the most convincing. An incoming dust transfer at least appears as something someone else sent you. The zero-value one appears as something you sent, with your address in the “from” column. If you remember sending to a counterparty recently and you see an outgoing line with a matching-looking destination, the natural reading is that this is your own transfer. Some interfaces show the amount as zero in small grey text, easy to miss when you’re scanning for an address rather than an amount.
Why multi-wallet farms are a soft target
Anyone can get poisoned. Farmers get more chances to make the mistake.
Think about the transfers a farm makes in a normal week. Gas goes out from a funding source to active wallets. Exchange withdrawals land in several addresses. Small leftover balances get swept. After a claim, tokens move toward wherever they’ll be held or sold. Each of those is a destination you’ll probably send to again, and every one is a pattern an attacker can copy.
Scale makes the history itself noisier. A wallet with three transactions a month is easy to scan, and an odd entry stands out. A wallet with sixty transfers across a campaign becomes a wall of truncated hex, and a poisoned line sits comfortably among dozens of real ones that look just as random.
Then there’s the mental state. Farming work tends to come in bursts: a gas window opens, a claim goes live, a bridge finally clears. Those are exactly the moments you’re moving fast, repeating the same step across many wallets, and least likely to stop and read an address properly. Attackers don’t need to know your schedule. The chain tells them when you’re active, because every transfer you make is public.
And farmers often keep many wallets open in one wallet app. If your interface lists twenty accounts, each with its own history, you have twenty places for a poisoned entry to sit and twenty chances to paste from the wrong one.
None of this means farming wallets are uniquely doomed. It means the copy step deserves a process, the same way funding and approvals already do.
The same address on many chains
There’s a wrinkle that catches people who farm across several EVM networks. An address derived from one key is the same string on every EVM chain. That’s convenient, and it also means a poisoning attempt doesn’t have to happen on the chain where you’ll make the mistake.
Say you regularly send gas to one wallet on three different rollups. An attacker who watches one of those chains can generate a lookalike of that destination and plant it there. The same lookalike address is valid on the other two chains as well. If your wallet shows a combined activity view across networks, the poisoned entry from one chain now sits next to your real transfers from all of them.
It works the other way too. Some wallets keep a separate history per network, so you may see the poisoned entry only on a chain you rarely open, and forget about it. Months later, you switch to that network to move funds, scroll history for the destination, and the fake is the most recent entry there.
Exchange deposit addresses add another layer. Many exchanges give you one deposit address per network, and those addresses often differ between networks even for the same asset. A farmer who withdraws and deposits across several networks ends up with a small zoo of exchange addresses that look nothing alike, which makes it harder to spot a stranger among them.
The practical point is to key your saved list by both address and chain, and to treat a destination as verified only on the network you actually verified it on. “This is my exchange deposit address” is an incomplete statement. “This is my exchange deposit address for this asset on this network, copied from the deposit screen on this date” is the version that holds up.
Why the wallet screen makes it hard to see
It’s tempting to blame the victim here and say they should just read the address. That’s fair as far as it goes, but interfaces genuinely work against you.
Truncation is the big one. When a wallet shows 0x7a3F…91cE, it’s telling you those are the parts that matter. They aren’t. The middle thirty-odd characters are where the real address lives, and they’re hidden by default in most lists.
Checksums don’t help either. Ethereum addresses can carry a mixed-case checksum that catches typos, where a single mistyped character makes the address invalid. A lookalike is a perfectly valid address that happens to belong to someone else, so every checksum check passes.
Identicons and colour blobs help a little. Some wallets and explorers draw a small pattern derived from the address. Two addresses with different middles usually produce different icons, and that’s a quick visual difference you can learn to notice. But people rarely remember what an icon looked like last week, so they catch the fake only when both addresses sit side by side.
Explorers have improved. Several now mark zero-value token transfers, hide or flag tokens from unverified contracts, or warn when an address looks similar to one you’ve interacted with. Some wallets filter spam transfers out of the default activity view. These features are worth turning on where they exist. They’re also inconsistent between chains and products, and you shouldn’t build a habit around a warning that might not fire on the next chain you use.
The test transaction trap
A common piece of advice is to send a small test amount first, confirm it arrived, then send the rest. That advice is sound, and address poisoning can turn it against you if you’re not careful about where the second address comes from.
Here’s the failure. You copy the correct address from your saved list, send a small test, and it confirms. An attacker watching the chain sees a new outgoing transfer to that destination and quickly fires a poisoned entry at your wallet, matching the destination you just used. By the time you come back to send the real amount, your history shows two recent entries that look alike: your test, and the fake. If you now copy “the address I just sent the test to” from history, you might pick the wrong one.
The fix is simple once you see it. The test proves the address in your saved source is correct. So send the main amount by copying from that saved source again, not from the test in your history. The test transaction should confirm a destination you already hold. It shouldn’t become the thing you copy from.
This is also why a quick test followed by a long pause is a bad pattern. The longer the gap between the test and the main send, the more likely a poisoned entry has landed in between, and the more likely you’ve forgotten exactly which line was yours. Do both sends in one sitting, from the same source.
Build an address book and treat history as untrusted
The strongest single habit is to stop using transaction history as a source of addresses at all. History is a record of what happened, including things other people made happen. It’s not a list you control.
An address book is a list you control. Most wallets have one: you save an address under a name, and when you send, you pick the name. Exchanges usually have a withdrawal address book too, often with an option to allow withdrawals only to saved addresses. Use both.
For a farm, the address book is also an operating record. I’d rather have a short, boring list with clear names than a wallet history I have to interpret. Something like “main vault, eth”, “gas source”, “exchange deposit, sol” beats remembering which truncated string was which. Keep the names functional. The point is to recognise a destination without reading hex.
If you also keep a list outside the wallet, keep the full addresses there, not shortened versions, and note which chain each belongs to. The same address can exist on several EVM chains, and a deposit address on one network may not be valid on another. The list should be the one place you copy from.
Two cautions. First, an address book is only as good as what you put in it. If you save an address by copying it from poisoned history, you’ve just saved the fake with a friendly name. Build the list from the original source: the exchange’s deposit screen, the wallet’s own receive screen, or the counterparty’s confirmed message. Second, protect the list. A plain text file on a shared machine is a target for the clipboard and file-reading malware covered in a separate piece on endpoint hygiene.
Verify the middle, not just the ends
Even with an address book, some sends will involve an address you’ve never saved: a new exchange deposit address, a new counterparty, a first-time bridge recipient. For those, change how you check.
Read characters from the middle. Pick a chunk somewhere around positions fifteen to twenty-five and compare it against the source. Attackers generate lookalikes to fool the view you’re given, which is almost always the ends. A middle chunk is the part they didn’t bother to match, because doing so costs far more work.
Compare against the original, not against memory. It sounds obvious, but many people check a pasted address by asking whether it “looks right”. Looking right is precisely what the attacker engineered. Put the source and the paste side by side.
Use the device screen when you have one. Hardware wallets show the full destination on their own display before you approve. That display is outside your computer, so it’s the most trustworthy view you have of where the funds are actually going. Reading the whole thing there, or at least a middle chunk and both ends, catches a swapped address whether it came from poisoning or from something worse on the computer itself.
If a transfer is large enough that losing it would really hurt, add friction on purpose. Read the address aloud in groups of four. Ask someone else to read it back. It feels excessive until the day it doesn’t.
Hiding and ignoring the poisoned entries
Once you know what poisoning looks like, the planted entries are easy enough to spot: a zero-value transfer you don’t remember, a dust amount from an address that almost matches a counterparty, a token with a familiar ticker from a contract you’ve never seen.
The right response is to do nothing with them. You don’t need to “clean” your history, and you can’t remove an event from the chain anyway. What you can do is hide them in your wallet’s view, where the wallet allows it, so they stop cluttering the list you scan. Some wallets let you hide specific tokens or turn on a spam filter. Explorers often have a toggle to hide zero-value transfers.
Don’t interact with the counterfeit tokens that come along with some poisoning attempts. Trying to sell or swap them is a separate trap, where the token itself or the site it points to is the attack. That’s covered in a separate piece on unsolicited tokens, and the short version is that anything you didn’t ask for is best left alone.
Don’t send anything back to the lookalike to “test” it or to complain. It won’t help, and it gives an attacker confirmation that your wallet is actively watched by a human.
It’s also reasonable to label the lookalike in your wallet or explorer if they support private labels. A note that says “poison, do not use” next to the fake address is a cheap backstop against a future tired version of yourself.
What to do if you already sent to a lookalike
This is the paragraph nobody wants to need. If you’ve sent funds to a poisoned address, they are almost certainly gone. On-chain transfers are final, and the recipient is a key the attacker controls.
There are still a few things worth doing, calmly. Confirm what happened by opening the transaction on an explorer and comparing the full destination against your real one. If the middle differs, you have your answer.
Check whether the asset is one where the issuer has freezing powers. Some centralised stablecoin issuers can freeze addresses, and they generally do so in response to formal reports rather than individual messages. Reporting through the issuer’s official channel won’t guarantee anything, but it’s the only route with a real chance of stopping further movement. Explorers also accept reports that let them label the address for other users.
If the funds went toward an exchange, that exchange’s compliance team is worth contacting through its official support system with the transaction hash. Exchanges receive a lot of these reports and outcomes vary, so treat it as a long shot.
Then comes the thing that matters most: ignore every “recovery service” that appears. Losses from address poisoning attract a second wave of people offering to trace and return funds for an upfront fee. Legitimate firms don’t approach victims in DMs or comment threads. Anyone who contacts you first with a recovery offer is running the next scam.
Finally, write down what happened and which habit failed. Did you copy from history? Check only the ends? Skip the address book because it was a new destination? That note is the part that actually prevents the next one.
What alternatives and shortcuts get wrong
A few shortcuts come up often. Some help a bit. None replace the basic habit.
Name services are the first. Sending to a human-readable name instead of hex removes the visual comparison problem, but it adds its own questions: whether the name resolves to what you expect on the chain you’re using, whether you typed the exact name, and what a public name reveals about which wallets belong together. That last point is covered separately. Names are useful, but they aren’t a cure.
QR codes are another. Scanning a code from the counterparty’s own receive screen is a good way to avoid pasting at all, provided the code comes from them directly. A QR code in a chat message or on a webpage is just a picture of an address, and it can be swapped like any other.
Spam filters come next. They reduce noise, and you should use them. But a filter that hides most poisoned entries teaches you to trust what’s left, and what’s left may include one it missed. A counterfeit token from a new contract might not be flagged yet.
Last is “I’ll just be careful”. Being careful is a mood. It holds up for a few days and fades during the first busy claim window. A process that doesn’t depend on your attention, like copying only from a saved list, holds up when you’re tired.
A Monday routine for a farm
If you run more than a handful of wallets, it’s worth spending an hour once to set this up properly. Here’s a version that fits most setups.
Build the master list. For every destination you send to repeatedly, capture the full address from its original source and record the chain, a functional name and the date you verified it. Keep it somewhere you trust, off the machine you browse on if you can.
Load the wallet address books from that list. Do the same for exchange withdrawal whitelists. Turn on the whitelist-only setting where the exchange supports it, knowing that most exchanges add a waiting period before a new address can receive withdrawals. That delay is a feature here.
Turn on whatever spam and zero-value filtering your wallets and explorers offer. Hide the counterfeit tokens you already have.
Set a rule for new destinations: verify a middle chunk against the source, send a small test if the amount justifies it, and send the main amount from the source again, not from history.
Then run a short check once a month. Scroll each active wallet’s history for poisoned entries. You’re not looking to clean them up. You’re checking whether you’re being targeted, because repeated poisoning attempts against your addresses are a sign your activity is visible and your patterns are being watched. That’s worth knowing when you think about how many wallets share the same destinations.
Where this advice stops working
I want to be straight about the limits. The habits here defeat address poisoning as it’s usually run: lookalikes matched to the visible ends of an address, planted in history. They don’t defend against everything that ends with funds at the wrong address.
If malware on your computer rewrites your clipboard, it can replace the address you copied from your perfect address book with one the attacker chose. Some clipboard hijackers even pick a lookalike so the swap survives a glance. The defence there is a clean machine and reading the destination on a hardware wallet screen, and that’s a separate topic.
If an attacker controls the source you copy from, such as a compromised exchange account or a message from a counterparty whose account was taken over, your address book will faithfully store the wrong address. Verifying new destinations through a second channel helps.
And there’s no fix for pure fatigue. Every one of these habits can be skipped at two in the morning. The best you can do is make the safe path the fast path: an address book entry is quicker to pick than scrolling history, once it exists.
I haven’t measured how often poisoning attempts hit farming wallets compared with ordinary ones, and I’m wary of anyone quoting a figure without showing how they got it. What’s clear from the mechanics is that more transfers give attackers more chances to try, and each of those tries depends on a paste you can choose not to make.
For more on wallet security, funding hygiene and the tools we actually use to run many wallets without cutting corners, head back to the Airdrop Farming homepage.
Get new guides and videos first — join the Telegram channel.