Address poisoning and fake airdrops turn a wallet's own transaction history into a trap, so a careful operator copies the wrong address or clicks a link that arrived as a "free" token. Neither scam needs a private key or a malicious approval to start: anyone can place a look-alike address or a token carrying a link into a public wallet's history. Academic researchers measured 270 million poisoning attempts against 17 million victims on Ethereum and BNB Smart Chain over two years, with at least $83.8 million lost across 6,633 incidents (USENIX Security 2025). This guide explains how the two scams work at the level a defender needs, then sets out address allowlisting, payment workflow controls, airdrop handling rules, detection and response for treasury teams, exchanges and the people who support their users. Approvals, permits and drainer kits are covered in the Crypto Scam and Drainer Guide; this page does not repeat them and does not describe how attack tooling is built.
Key takeaways
- History is not an address book. Any address that appears in a wallet's history or a block explorer may have been planted there by a stranger, including entries that look like your own addresses.
- First and last characters prove nothing. Attackers generate vanity addresses that match the visible start and end of a real counterparty, so truncated displays and quick visual checks fail by design.
- Allowlists beat vigilance. A destination allowlist with a time delay and dual approval for changes removes the copy and paste step that poisoning depends on.
- An unsolicited token is a message, not a gift. Fake airdrops carry links in token names or memo fields that lead to wallet connection or seed phrase capture; the safe default is to ignore them entirely.
- Losses are usually final. On-chain transfers cannot be reversed, so prevention, fast reporting with transaction IDs and refusal of "recovery" offers are the only levers left.
How address poisoning works
Poisoning exploits a habit: copying a recent destination from transaction history instead of from a verified record. MetaMask's support guidance says the scam "relies on user carelessness and haste above all else." The attack has three parts.
- Watching: attackers monitor public transfers, often of stablecoins, between a wallet and its regular counterparties. Etherscan notes that spoofed transfers often appear within a minute of a large stablecoin transfer.
- Look-alike generation: they generate a vanity address whose first and last characters match the counterparty (or the victim's own address), so it looks identical when a wallet or explorer truncates it.
- Planting: they make that look-alike appear in the victim's history, then wait for the next payment where someone copies the wrong line.
Etherscan describes three planting methods, summarised below. OpenZeppelin explains why the zero-value variant works: the ERC-20 transferFrom function only checks that the amount does not exceed the allowance, and since the default allowance is zero, a zero-value call between two arbitrary addresses passes and emits a normal Transfer event that wallets and explorers then index.
| Variant | What appears in the victim's history | Why it fools people | Primary counter |
|---|---|---|---|
| Zero-value transfer (event spoofing) | A transfer of 0 of a real token, "from" the victim to a look-alike address | It looks like an outgoing payment the victim made themselves | Hide zero-value transfers; never copy from history |
| Dust transfer from look-alike | A tiny incoming amount from a look-alike of a known counterparty | The sender looks like a regular counterparty | Allowlisted destinations only |
| Counterfeit token transfer | A token with the same name and symbol as USDT or USDC, sometimes mirroring the real payment amount | Amount, symbol and timing all match a genuine payment | Token reputation filtering; check contract address, not symbol |
Trezor's support guidance notes three limits worth stating to users: poisoning cannot drain a wallet by itself, it does not involve a private key leak, and it targets low-fee EVM chains where mass sending is cheap. The damage comes from a single transfer to the wrong destination. Because the victim signs a perfectly valid transaction, security tooling that looks for malicious contract calls sees nothing unusual.
Why checksums and truncated displays do not help
Teams sometimes assume address checksums catch mistakes. ERC-55 (status: Final) encodes a checksum in the upper and lower case of an Ethereum address so that a random typo has, on average, a 0.0247% chance of passing. That protects against mistyping, not substitution: an attacker's look-alike is a valid address with its own valid checksum, so it passes every format check. The same applies to Bitcoin address checksums.
Display choices matter more. Wallets and explorers that show 0x1234...abcd hide exactly the characters an attacker could not match. Etherscan has responded by showing more leading and trailing characters, highlighting matching addresses on hover and adding a copy-address warning on token transfer pages. Internally, finance and operations tools should show full addresses or verified labels, never the truncated form.
Human-readable names (for example Ethereum Name Service records) reduce copy and paste errors but move the risk to name spoofing. Etherscan explicitly warns that domain names can be spoofed too. Treat a name as a convenience that must resolve to an address already on your allowlist.
How fake airdrop scams work
A fake airdrop uses the same public delivery route. Anyone can send a token or NFT to any address, and the token itself becomes the lure. The FBI's June 2025 public service announcement (I-060325-PSA) on Hedera network wallets describes two patterns that generalise to other chains:
- Memo or token-name lure: an unsolicited token arrives with a plaintext memo or a name containing a URL. The site asks the user to connect a wallet to a decentralised app, enter a password or type a seed phrase.
- Off-chain lure: fake airdrop offers spread through social media, look-alike project sites and phishing emails, often imitating real projects that have announced or are rumoured to announce a token.
The end of the flow is one of three things: a seed phrase capture (total loss), a malicious signature or approval (covered in the drainer guide), or a "claim fee" paid to the scammer. A related pattern sends worthless tokens that cannot be sold, prompting the holder to visit a "swap" or "unlock" site. The spoofed Transfer events Etherscan describes can also be used to make it look as if a well-known address has just sent tokens or that an airdrop is under way.
| Signal | Legitimate airdrop | Likely scam |
|---|---|---|
| How you heard about it | The project's own verified channels, announced in advance | A token that appeared unasked, a DM, or a reply under a project's post |
| What it asks for | A claim on the project's known domain, sometimes nothing (tokens just arrive) | A seed phrase, password, "verification" signature or fee |
| Where the link came from | Typed or bookmarked domain | Token name, memo field, NFT image text or shortened URL |
| Urgency | Published claim window | Countdowns, "last chance", limited spots |
| Token contract | Matches the address the project publishes | Unverified contract sharing a well-known name or symbol |
No legitimate wallet, exchange or project needs a seed phrase to deliver a token. That single rule stops the most damaging version of the scam.
Controls for treasury and operations wallets
Organisations that move crypto (treasury teams, market makers, payment processors, protocol foundations) should design payments so that nobody ever copies a destination from history. The control set below assumes a multisig, MPC or custodial platform; see the Multisig Treasury Operations Guide and the MPC Wallet Security Guide for the signing side.
- Destination allowlist: outgoing transfers only to addresses on an approved list, enforced by the custody or signing platform rather than by policy alone. Most institutional custody and multisig platforms support an address book or whitelist policy; use the enforcing version, not a reference list.
- Out-of-band onboarding: add a new counterparty address only after confirming it through a second channel already on file (a known phone number or a signed message from the counterparty's existing address), never from the email or chat that requested the payment.
- Change delay and dual control: new or edited allowlist entries take effect only after a cooling-off period set in policy and need approval from someone other than the requester.
- Test transfer for new destinations: send a small amount, have the counterparty confirm receipt and the exact amount, then send the balance. Trezor notes the trade-off: this costs fees twice.
- Full address verification on the signing device: approvers read the whole address on the hardware wallet or signing screen against the allowlist record, not against the request.
- Spam hygiene in tooling: turn on zero-value and low-reputation token filtering in explorers and wallet software used by finance staff, and label known counterparties with private name tags.
Small organisations that cannot run a custody platform can still get most of the value: keep counterparties in the wallet's contacts list, require a second person to check the full address for any payment above a set threshold, and keep operating balances small. Large organisations should add automated pre-signing checks that compare every destination with the allowlist and with recently seen look-alikes.
Controls for exchanges and wallet providers
Exchanges and wallet providers carry most user-facing exposure: customers withdraw to self-custody addresses they copied from somewhere, and poisoned look-alikes target exactly that step. MetaMask and Trezor both warn that users should not copy even their own address from history when moving funds from an exchange.
| Control | What it does | Who it suits |
|---|---|---|
| Withdrawal address allowlist with lock period | Withdrawals only to saved addresses; new entries usable after a delay and an email or app confirmation | All exchanges; default on for high-value accounts |
| Look-alike detection at send time | Compares the destination with history and warns or blocks when start and end match a past address but the middle differs (MetaMask describes this as a blocking warning) | Wallets and exchange withdrawal flows |
| First-time destination warning | Flags any address the account has never paid before | Wallets and exchanges |
| Zero-value and spam token hiding | Hides zero-value transfers and low-reputation tokens by default, with an opt-in toggle (OpenZeppelin recommendation) | Wallets, explorers, exchange deposit history |
| No history-prefilled address fields | Address fields never auto-suggest from history | Wallet and exchange UI teams |
| Airdrop quarantine | Unsolicited tokens hidden or marked unverified; links in token names and memos not rendered as clickable | Wallets and portfolio views |
Support teams need scripts for both scams: a customer who has just sent funds to a look-alike needs the transaction hash, an immediate report route and a clear warning about recovery scams, not a promise to reverse the transfer. Where the asset is a stablecoin, the issuer may have freezing capability; see the Stablecoin Security and Risk Guide for how issuer controls work and why speed matters. Exchange account protections more broadly are covered in the Crypto Exchange Account Security Guide.
Rolling out the controls
- Inventory payment paths: list every wallet, custody account and exchange account that can send funds, who can initiate and approve, and where destination addresses currently come from.
- Build the allowlist in report mode: export the last 6 to 12 months of outgoing destinations, verify each one with the counterparty out of band, and remove anything unverified. Run the platform's allowlist policy in alert-only mode for two weeks to see which payments would have been blocked.
- Enforce for high-value wallets first: switch treasury and hot wallets with the largest balances to enforced allowlisting with change delays; keep a documented break-glass route that needs two senior approvers.
- Clean the tooling: enable zero-value hiding, token ignore lists and name tags in explorers; set wallet software to warn on first-time and look-alike destinations.
- Train with real examples: show finance staff screenshots of poisoned histories and fake airdrop memos from your own wallets. Make "ignore unsolicited tokens" and "never type a seed phrase into a website" explicit policy.
- Extend to customers: exchanges and wallets roll out allowlisting defaults, look-alike warnings and airdrop quarantine in stages, measuring withdrawal friction and support contacts.
- Roll back safely: if enforcement blocks a legitimate urgent payment, use the break-glass route and record it rather than disabling the policy globally.
Detection and measurement
What to monitor:
- Inbound zero-value transfers and counterfeit-token transfers to corporate wallets, especially from addresses whose first and last characters match a known counterparty (these are attempts against your staff).
- Any outgoing transfer, or attempted transfer, to a non-allowlisted destination.
- Allowlist change requests, who raised them and through which channel.
- Unsolicited tokens and NFTs arriving in corporate wallets, and any staff interaction with them.
- Look-alike domains and social accounts impersonating your brand around token events.
| Metric | Target direction | What it shows |
|---|---|---|
| Share of outgoing value sent to enforced-allowlist destinations | Towards 100% | Whether poisoning can still reach a payment |
| Allowlist changes verified out of band within policy | 100% | Onboarding discipline |
| Poisoning attempts detected against corporate wallets per month | Track, not reduce | Exposure; spikes follow large public transfers |
| Look-alike send warnings overridden by users | Down | Whether warnings are understood or ignored |
| Median time from loss report to issuer and law enforcement notification | Down | Chance of freezing funds |
Responding to a poisoned transfer or airdrop compromise
- Confirm and record: capture the transaction hash, source and destination addresses, asset, amount, timestamps and how the address was obtained.
- Contain: if a seed phrase was typed or a wallet connected to a fake airdrop site, move remaining assets to a fresh wallet created on a clean device and revoke any approvals. If only an address was copied, check that no other payments used the same source.
- Notify quickly: contact the stablecoin issuer or the receiving exchange if funds reached one, and report to law enforcement. In the US the FBI asks victims to report to IC3 with wallet addresses, amounts, dates, times and transaction IDs; elsewhere use the national cybercrime or fraud reporting service.
- Refuse recovery offers: the FBI warns that people offering to recover lost funds may be running a second scam.
- Fix the process: find where the address was copied from and remove that path, then add the counterparty correctly to the allowlist.
The Cryptocurrency Scam Awareness Guide gives consumer-facing wording that support teams can reuse.
Common failure patterns
- Allowlist as a spreadsheet: a list nobody enforces; staff still copy from the explorer under time pressure.
- Onboarding from the request channel: a "new address" arrives by email or chat and is added without a callback, combining poisoning with business email compromise.
- Checking only the ends: approvers compare four leading and four trailing characters, exactly what vanity generation matches.
- Trusting the token symbol: a counterfeit "USDT" transfer for the same amount as a genuine payment is accepted as proof of the counterparty's address.
- Curiosity about airdrops: a staff member opens a link from an unsolicited token on a treasury wallet's portfolio page and connects the wallet.
- Self-transfer shortcuts: moving funds between your own exchange and cold wallet by copying "your" address from history.
Framework mapping
| Control area | NIST CSF 2.0 | NIST SP 800-53 Rev 5 | ISO/IEC 27001:2022 Annex A |
|---|---|---|---|
| Staff awareness of poisoning and airdrop lures | PR.AT-01, PR.AT-02 | AT-2 | 6.3 Information security awareness, education and training |
| Enforced destination allowlist and dual approval | PR.AA-05 | AC-6 | 5.15 Access control |
| Monitoring of wallets and look-alike transfers | DE.CM-09, DE.AE-02 | SI-4, AU-6 | 8.16 Monitoring activities |
| Loss response and reporting | RS.MI-01 | IR-4 | 5.26 Response to information security incidents |
Practitioner checklist
- Enforce destination allowlisting on every treasury, hot and exchange account that can send funds.
- Verify every new counterparty address out of band and apply a change delay with dual approval.
- Ban copying destinations from transaction history or explorers, including your own addresses.
- Require full-address checks on the signing device and a test transfer for first-time destinations.
- Enable zero-value hiding, token ignore lists and name tags in tools finance staff use.
- Treat every unsolicited token or NFT as a phishing message: never click, connect or claim.
- State in policy that no legitimate party ever needs a seed phrase.
- Monitor inbound look-alike transfers and alert on any non-allowlisted outgoing attempt.
- Keep a loss runbook with issuer, exchange and law enforcement contacts and a recovery-scam warning.
Standards and references
- FBI IC3: Cyber Criminals Defraud Hedera Hashgraph Network Non-Custodial Wallet Users Through Nonfungible Token Airdrops
- Tsuchiya, Dong, Soska and Christin: Blockchain Address Poisoning (USENIX Security 2025)
- Ethereum Improvement Proposals: ERC-55 Mixed-case checksum address encoding
- Etherscan: What is address poisoning
- OpenZeppelin: How to ensure Web3 users are safe from zero transfer attacks
- MetaMask: Address poisoning scams
- Trezor: What are address poisoning attacks and how to avoid them
Related NHI Mgmt Group resources: Crypto Scam and Drainer Guide, Crypto Wallet Security Guide, Multisig Treasury Operations Guide, Digital Asset Custody, Authorised Push Payment (APP) Fraud Guide.