Address Poisoning in 2026: The Copy-Paste Crypto Heist

On May 3, 2024, someone tried to send roughly $68 million in Wrapped Bitcoin to a wallet they had paid before. They did what almost everyone does with a repeat payment: they opened a recent transaction, copied the recipient address, pasted it, and hit send. The first characters matched. The last characters matched. The long run of characters in the middle, the part a wallet politely hides behind an ellipsis, did not. The funds landed with a stranger whose total outlay for the theft was about 65 cents in gas, according to the Carnegie Mellon University researchers who later reconstructed the operation.

That transfer is the single most expensive case of a scam called address poisoning, and it was not bad luck. Address poisoning is a patient, automated, industrial fraud that turns your own transaction history into the bait. There is no malicious contract to approve, no drainer signature to sign, no fake website to visit. There is only a lookalike address, sitting quietly in your activity feed, waiting for one distracted copy-paste. This is how the attack works, how large it has quietly become, why code audits do nothing to stop it, and what actually protects a wallet in 2026.

None of that requires a hacked exchange, a leaked seed phrase, or a zero-day in a smart contract. It requires only that a busy person, at the wrong moment, trusts a line in their own wallet. That is what makes address poisoning both the most low-tech and, measured against the fraction of a cent it costs to try, one of the most efficient thefts in crypto, and why a technique that drains tens of millions of dollars a year keeps outrunning every attempt to stamp it out.

What address poisoning actually is

Most crypto theft in the headlines involves getting a victim to authorize something: signing a token approval, blind-signing a transaction, or handing a seed phrase to a fake support agent. Address poisoning is different, and in a way more unsettling, because the victim authorizes nothing to the attacker directly. They simply send their own money to the wrong address, and they do it themselves, with a clear head, on a wallet that has not been compromised in any technical sense.

The mechanics rest on a single human habit. Wallet addresses are 42-character hexadecimal strings that no one reads in full. To send a repeat payment, people copy an address from somewhere they trust, and the most convenient source is their own transaction history. The attacker’s job is to plant an address in that history which looks, at a glance, exactly like the one the victim means to use. When the victim copies the wrong line, the transfer is irreversible and the chain does exactly what it was told.

It is, at heart, a phishing attack, but the lure is not an email or a spoofed domain. The lure is a row in a ledger the victim already believes. Microsoft’s security team popularized the term ice phishing for on-chain approval scams back in 2022; address poisoning is the colder cousin, because it never asks for a signature at all.

The distinction matters for defense. When a drainer empties a wallet, the lesson is not to sign what you do not understand. When a support scammer extracts a seed phrase, the lesson is that no one legitimate ever asks for it. Address poisoning offers no such tidy rule about whom to trust, because the victim was not tricked into trusting the attacker at all. They were tricked into trusting themselves, or rather their own record of what they had already done. Defeating it means distrusting a source almost no one thinks to question.

Why the two addresses look identical

The attack only works because of how software displays addresses. A full Ethereum address looks like 0x71C7656EC7ab88b098defB751B7401B5f6d8976F, and every wallet and block explorer shortens it for readability, showing something like 0x71C7…976F. Users learn to recognize the short form and stop checking the middle, which is exactly the shortcut the attacker is counting on.

Attackers exploit that habit with vanity address generators, brute-force tools that grind through candidate keys until they find one whose address happens to reproduce the first few and last few characters of the target. Matching four leading and four trailing hex characters is cheap; matching more gets exponentially harder, but attackers only need to beat the glance, not the full string. Blockaid, a transaction-security firm, has documented campaigns using specialized generators that reproduce the first and last several characters of a legitimate address.

The economics are brutal for the defender. Generating a lookalike costs a few seconds of GPU time. Seeding it into a history costs a fraction of a cent in gas. The payoff, if the victim is a whale, runs to eight figures. That asymmetry is why the technique scaled from a curiosity into one of the most prolific frauds on-chain.

There is a probabilistic arms race buried in that display. An Ethereum address carries 40 hexadecimal characters, so matching any fixed set of them is a needle-in-a-haystack search, but graphics hardware makes billions of guesses cheap. Reproducing the leading four and trailing four characters, the slice most interfaces show, takes seconds; reproducing six or seven at each end is still within reach of a patient operator. Since hardly anyone compares more than the ends, the attacker rarely needs to aim higher.

Three ways to poison a history

In the most thorough study of the technique to date, Carnegie Mellon researchers Taro Tsuchiya, Jin-Dong Dong, Kyle Soska and Nicolas Christin catalogued three distinct ways attackers get a lookalike address into a victim’s feed. Each abuses a different quirk of how tokens and explorers behave, and all three share one goal: put a poisoned line next to a legitimate one and wait.

The three variants matter because they defeat different instincts. Dusting fools the user who checks who paid them; the counterfeit-token trick fools the user who checks the amount; and the zero-value transfer, described below, fools the user who trusts an outbound entry because it looks like something they did themselves.

One trait unites all three. None of them touches the victim’s keys or balances at the moment of seeding; the poisoning transaction is just data, a log entry built to be read by a human days or weeks later. The theft itself happens entirely inside the victim’s own later transfer, which is why nothing in a standard security setup raises an alarm while the trap is being laid.

The zero-value transfer trick, in plain English

The zero-value variant deserves a closer look, because it surprises even experienced users. Standard ERC-20 tokens let anyone call transferFrom to move tokens between two addresses, provided the caller has an allowance. The catch is that moving zero tokens needs no allowance at all, and many token contracts still emit a Transfer event for it.

That event is the whole game. Block explorers and wallets read Transfer events to build your activity feed. When the attacker fires a zero-value transferFrom that records your address as the sender and the lookalike as the recipient, the explorer dutifully lists an outbound transfer that you appear to have made. To a user scanning their history for the last address they paid, it reads like a prior payment to a familiar destination. It is not; it is a log entry the attacker minted for a fraction of a cent, with no tokens ever changing hands.

Etherscan and other explorers have since started hiding or flagging zero-value transfers precisely because they carry no economic meaning and exist only to deceive. But the underlying token standard still allows them, and not every interface filters them out, so the trick keeps working wherever a wallet shows raw Transfer events without judgement.

Interfaces have made this harder in fits and starts. Some now suppress zero-value transfers by default, and others label incoming transfers from unknown contracts as suspected spam. But coverage is uneven across chains and clients, and attackers simply migrate to whichever variant a given interface still renders cleanly. The counterfeit-token version is especially durable, because a wallet cannot always distinguish a fake USDT contract from the genuine one without extra metadata it may not have.

270 million attempts: the scale nobody sees

Because each individual attempt costs almost nothing, attackers spray poisoned transactions at a volume that is hard to picture. The Carnegie Mellon team built a detection system that ran across Ethereum and BNB Chain from July 1, 2022 to June 30, 2024 and counted roughly 270 million on-chain poisoning attempts aimed at about 17 million victim addresses, 13 times more than any prior estimate. Of those, they confirmed 6,633 successful thefts totaling at least $83.8 million. In their words, that makes “blockchain address poisoning one of the largest cryptocurrency phishing schemes observed in the wild,” a conclusion set out in their research paper and presented at USENIX Security.

The numbers have not cooled since. Blockaid reported more than 65.4 million flagged poisoning transactions since January 2025 and roughly 316,000 confirmed successful attacks, about one in every two hundred attempts. The gap between attempts and hits is the entire business model.

Hundreds of millions of poisoned lines produce only a few thousand catastrophic hits, but each hit can be enormous, so the expected value stays firmly positive for the attacker. It is spam economics applied to theft.

That volume is only possible because the marginal cost of an attempt rounds to nothing. A poisoning transaction is a single token transfer, often batched by the thousand, and on cheap chains or in quiet blocks it can cost a tiny fraction of a cent. The attacker does not need the hit rate a targeted spear-phishing campaign would demand; they need one better than the near-zero cost of trying, and against seventeen million wallets even a vanishingly small success rate yields thousands of paydays.

Anatomy of the $68 million heist

The May 2024 WBTC theft shows the model at full stretch. According to Chainalysis, which reconstructed the campaign, the operation ran for 66 days, from February 28 to May 4, 2024, and seeded 82,031 lookalike addresses, a bit under one percent of all new Ethereum addresses created in that window.

Most of that effort caught nothing. Of the tens of thousands of poisoned addresses, only 22 received more than $100 from a genuine victim. But the 2,774 addresses that did get funded took in a combined $69.7 million, because the operation was aimed at wallets with real balances. The average targeted wallet held about $338,900 and had made nearly 600 transactions across more than 500 days of activity. These were not naive newcomers; they were busy, experienced users moving large sums, exactly the population most likely to paste from history on autopilot. Experience is not protection here, it is the vulnerability, because familiarity is what turns verification into a reflex people skip.

The single $68 million transfer, first reported as it happened in May 2024, is what made the campaign infamous, and it had an unusual coda. After days of on-chain negotiation and public pressure, the thief returned the funds, keeping roughly $3 million and handing back the rest in early May 2024. That outcome is the exception, not the rule; most poisoned transfers are simply gone. The forensics a security team can reconstruct after the fact do nothing for the victim in the moment, which is why every serious defense against this attack is preventive.

The return of the money was extraordinary, and it should not be read as reassurance. It happened because the sum was large enough to summon intense public scrutiny, professional negotiators, and blockchain-intelligence firms racing to trace the flow, and because the thief judged that quietly keeping a few million beat holding sixty-eight million under a spotlight. A victim who loses a hundred thousand dollars gets none of that machinery. For them the transfer is final the instant it confirms.

The whale-hunting turn: $50 million and $12.4 million

If 2024 proved the technique could land a nine-figure hit, 2025 and 2026 proved it had become a specialty in hunting large wallets rather than spraying small ones. This mirrors a broader shift across crypto phishing that HOGE Wire has tracked as “fewer nets, bigger fish”: mass-market drainer losses have fallen while a handful of targeted operations against whales set records.

Two incidents bracket the trend. In December 2025, a single victim lost close to $50 million in Tether’s USDT after copying a poisoned address; Blockaid noted the funds went out just 26 minutes after the attacker detected a test transaction, suggesting the target was being watched in real time. Weeks later, on January 30, 2026, another victim sent 4,556 ETH, about $12.4 million, to a lookalike that had been sitting in their history after a two-month dusting campaign.

According to Scam Sniffer data reported in February 2026, Ethereum address poisoning alone drained about $62 million across those two months, even as the total number of victims stayed small. A contributing factor is mundane: cheaper block space. As base-layer and layer-2 fees have fallen, seeding thousands of poisoned transactions costs less than ever, so campaigns cast wider for the same budget while concentrating their real hopes on a few high-balance targets.

The real-time element is what should worry large holders most. In the December case the attacker was not passively waiting; they were watching the target’s wallet closely enough to notice a test transaction and to anticipate that a full transfer would follow within the hour. That is surveillance, not luck, and it points to operators who study their marks the way a targeted intrusion team would, then let a poisoned address do the actual work.

It is not just an Ethereum problem

Address poisoning is a property of how humans read addresses, not of any one chain, so it appears everywhere addresses are long and truncated. On Tron it targets USDT senders; on BNB Chain it made up half of the Carnegie Mellon dataset; on Solana and other networks the same lookalike-and-dust pattern shows up.

Even Bitcoin, whose addresses were once thought too structurally distinct to spoof convincingly, is now a target. Jameson Lopp, chief security officer at the custody firm Casa, documented roughly 48,000 Bitcoin poisoning transactions over 18 months starting in mid-2023, and urged holders to verify full addresses and wallet makers to build interfaces that display addresses in full rather than truncating them. His point cuts to the root of the problem: the ellipsis is not a convenience, it is an attack surface.

A newer wrinkle turns smart-contract wallets against their owners. Blockaid described a campaign that generated around 15,000 malicious proxy addresses tied to Safe-style nested proxies, dressing up attacker-controlled contracts to resemble legitimate multisig deployments. As more value moves through programmable wallets, the surface for this kind of impersonation grows, a theme that runs through the broader shift to account abstraction.

The cross-chain spread also complicates cleanup. A poisoned address on one network reveals nothing about the same crew’s footprint on another, and stablecoins like USDT, which move across many chains, hand poisoners a consistent, high-value target wherever they operate. Tron, where USDT transfers are cheap and constant, has become especially fertile ground for the dusting variant.

Where it fits in the phishing economy

Address poisoning does not exist in isolation; it is one product line in a large, professional scam economy. Chainalysis has estimated that crypto scams generated on the order of $17 billion in revenue in 2025, with impersonation scams rising sharply year over year, in its 2026 crime report. Poisoning sits alongside wallet drainers, approval and signature phishing, pig-butchering, and outright exit scams as interchangeable ways to separate a holder from their coins.

The supply side is industrialized. Chainalysis has documented plug-and-play poisoning toolkits sold on darknet markets, bundling lookalike-address generators, automated seeding scripts, and step-by-step guides, the same as-a-service model that powers drainer kits and the rug-pull assembly lines HOGE Wire has covered elsewhere. Once the funds are stolen, they enter the same laundering pipelines as any other illicit crypto, moving through mixers, cross-chain bridges and swap services that sit in the blind spots regulators are still trying to close.

It also rhymes with the signature-phishing family, where the exploit is not a stolen key but a victim clicking approve on something they do not understand. New account-delegation features have widened that surface further; the same social engineering that plants a poisoned address can push a victim toward a malicious delegation under standards like EIP-7702. Attackers treat all of these as one menu, choosing whichever lure fits the target.

What binds the economy together is fungible infrastructure. The same wallet-clustering that helps a drainer crew find rich targets helps a poisoning crew choose which addresses to seed; the same cash-out brokers move the proceeds; the same channels sell the kits. Treating address poisoning as an isolated trick misses the point. It is one feature on the menu of organizations that run every other flavor of on-chain fraud, switched on whenever a target looks likelier to paste than to sign.

Why smart contracts cannot save you

A security audit reads code. Address poisoning has no code to read, at least not in the victim’s transaction. The token contract behaves exactly as written; the wallet signs exactly what the user tells it to; the network settles a valid transfer to a valid address. Nothing is exploited except the gap between the address the user sees and the address they mean.

That is why the fix cannot come from smart-contract security, which is built to stop a contract from doing something it should not. Here the contract does precisely what it should. The failure is at the interface: the ellipsis that hides the middle of an address, the activity feed that treats a poisoned line the same as a real one, and the muscle memory of copy-paste. Defenses have to live at that display and workflow layer, in wallets, explorers and user habits, not in a Solidity audit.

This is also why the attack survives every advance in on-chain security. Better key management, hardware wallets, multisig, and clear-signing all address the problem of authorizing the wrong action. Address poisoning gets the user to authorize the right kind of action to the wrong destination, and none of those tools inspect whether the destination is the one the user intended. Solving it takes a different discipline: checking where money is going, not just what it is doing.

Precision about where responsibility sits matters, because the framing decides the fix. The token standard is not broken; permissionless transfers and their events are doing exactly what they were designed to do. The wallet is not broken; it is faithfully displaying real on-chain data. The soft spot is a design decision, made years ago and copied everywhere, to abbreviate addresses for readability and to present all history as equally trustworthy. That is a product choice, and product choices can be revisited.

The tell-tale signs of a poisoned history

Because address poisoning hides in plain sight, the first line of defense is learning to recognize a poisoned entry. None of the signs need special software, only the habit of pausing before a large or routine send and asking whether a line in your history is really what it seems.

• An unexpected tiny or zero-value transfer you never initiated, especially one that mirrors a payment you just made.
• A recent entry whose counterparty shares your real recipient’s first and last characters but differs somewhere in the middle.
• An inbound token that resembles a stablecoin you hold yet arrived from an address you do not recognize.
• A transfer showing exactly zero tokens moved, or a token whose contract address is not the official one.
• Any new address that surfaces in your feed right after a large or habitual transaction, which is precisely when attackers time their seeding.

The common thread is timing and mimicry. The poisoned line always lands close to a genuine one and always resembles it, because its whole purpose is to be mistaken for the real thing at a glance. Once you expect that pattern, the trap is far easier to sidestep than to fall into.

The defense stack: how not to get poisoned

The good news is that a handful of habits defeat the attack almost completely, because the attacker’s entire edge is a moment of inattention. The single most important rule, repeated by every wallet and analytics firm that has studied this, is to never copy a recipient address from your transaction history.

Address books deserve special emphasis because they are the cleanest fix: once a counterparty is saved, the poisoned line in your history becomes irrelevant, since you never touch it again. Ledger’s security team and others also recommend a test transaction for any large or first-time transfer, a small cost that caps the downside of a mistake. Name services go one step further by taking the raw address out of the human’s hands altogether.

What wallets, explorers, and regulators are doing

The ecosystem has been slowly hardening the display layer the attack depends on. MetaMask has warned users directly about address poisoning, and Etherscan, whose truncated display helped enable the scam, now describes the attacks as increasingly automated and widespread and hides zero-value transfers that exist only to deceive. Coinbase Wallet, Binance and others have added address-book features and spam-address detection to keep poisoned entries out of the way.

Detection firms have made it their business. Raz Niv, co-founder and chief technology officer of Blockaid, frames the company’s role as protecting users from malicious dApps, drainers, address poisoning and suspicious transactions before they sign, using transaction simulation that previews the true destination of funds rather than trusting the displayed address. Lopp’s call for wallets to stop truncating addresses points at the same deeper fix from the user’s side: an interface that never hides the part of the address the attacker is exploiting.

Regulators treat poisoning as part of the broader fraud problem rather than a distinct offense. The FBI’s Internet Crime Complaint Center logged $9.3 billion in crypto-related losses in its 2024 report, and the SEC routinely warns retail investors about impersonation and social-engineering scams. But recovery is rare and prosecution rarer, because the transfer is technically valid and the attacker is usually anonymous and offshore. The uneasy question of whether draining a wallet through pure deception is even a chargeable on-chain crime echoes the debates HOGE Wire has followed in cases like the Mango Markets appeal.

A coordination problem sits underneath all of this. No single wallet or explorer can fix the habit alone, because a user poisoned in one interface can still act on the bad address in another. Real progress depends on the whole stack, wallets, explorers, name services, and hardware makers, settling on defaults that make the full destination hard to ignore and easy to check. Until that happens, the burden falls back onto the user, which is exactly where a social-engineering attack wants it.

The outlook: cheaper dust, patient money

Everything that makes crypto cheaper and faster also makes address poisoning cheaper and faster. Falling fees on Ethereum and its layer-2s cut the cost of seeding poisoned transactions; better analytics make it easier to identify high-value wallets and time an attack to a victim’s real-time activity. The December 2025 heist, where funds moved 26 minutes after a test transaction, is a preview of campaigns that watch their targets and strike on cue.

There is a more hopeful reading, too. Address poisoning is unusually solvable for a fraud of its scale, because its single point of failure sits with the defender rather than the attacker. Every other lever, cheaper dust, sharper targeting, faster surveillance, only helps the attacker at the margins; none of them can force a transfer if the sender refuses to reuse an address from history. The cure is unglamorous and entirely within reach, which is both the frustration and the promise.

None of that changes the defense, because the attack has one dependency: a human copying an address they did not fully verify. That dependency has one cure. Never trust the history, always verify the destination, and use an address book or a name for anyone you pay more than once. Jameson Lopp’s blunt advice for Bitcoiners generalizes to every chain: check the address, and demand wallets that show you all of it.

Address poisoning will not produce the dramatic headlines of a bridge exploit or an exchange breach, because there is no breach; there is only a ledger telling the truth about a mistake. That is exactly what makes it dangerous, and exactly why the people least likely to fall for it are the ones who assume they already have.

Frequently Asked Questions

What is address poisoning in crypto?

Address poisoning is a scam where an attacker plants a lookalike wallet address into your transaction history, usually by sending a tiny or zero-value transfer from an address that matches the first and last characters of one you use. The goal is to get you to copy the wrong address from your history and send funds to the attacker instead of your intended recipient.

How do attackers make an address look like mine?

They use vanity address generators that brute-force candidate addresses until one matches the first few and last few characters of the target address. Because wallets and explorers shorten addresses to something like 0x71C7…976F, matching only the visible ends is enough to fool a quick glance, even though the middle of the address is completely different.

Can I get my funds back after an address poisoning attack?

Almost never. The transfer is a valid, irreversible transaction to an address you authorized, so there is no protocol-level way to reverse it. The famous $68 million case in May 2024 was returned only because the thief chose to send it back after public pressure; the vast majority of poisoned transfers are permanently lost.

How do I protect myself from address poisoning?

Never copy a recipient address from your transaction history. Save verified addresses in an address book or use a name service so you select a counterparty by name, send a small test transaction before any large transfer, verify the full address rather than just the first and last characters, and confirm the destination on your hardware wallet screen.

Is address poisoning only a problem on Ethereum?

No. It affects any chain with long, truncated addresses, including BNB Chain, Tron, Solana and even Bitcoin. Carnegie Mellon researchers found hundreds of millions of attempts across Ethereum and BNB Chain, and Casa’s Jameson Lopp documented tens of thousands of Bitcoin poisoning transactions, so the same precautions apply everywhere.

By Anneke de Vries, senior security correspondent at HOGE Wire.