Threat reportPhishingTL-2026-0837
Approval Phishing: Wallet-Draining Crypto Scams via Malicious Token Approvals (approve/permit/setApprovalForAll)
Approval phishing tricks a victim into signing what looks like a minor wallet interaction that actually grants an attacker-controlled address an on-chain spending allowance, letting the scammer drain the wallet's tokens at will. On-chain scams reached at least $14 billion (likely $17 billion) in 2025, with the average payment to a single scam address up 253% year-over-year and AI-augmented scams 4.5x more profitable.
Key facts
| Identifiers | TL-2026-0837 |
|---|---|
| Also known as | Approval phishing, Token approval phishing, Wallet drainer, ice phishing, Permit phishing, setApprovalForAll phishing |
| Severity | HIGH |
| Affected products | Ethereum / EVM ecosystem ERC-20 token allowance model (approve / increaseAllowance); Ethereum / EVM ecosystem ERC-721 / ERC-1155 NFT operator approvals (setApprovalForAll); Tether USDT (ERC-20 and TRC-20) |
| Weaknesses (CWE) | CWE-1021, CWE-862, CWE-451 |
| Attribution | Pig-butchering (medium confidence) |
| Malware | Pink Drainer |
| Tools | Angel Drainer, Inferno Drainer, Wallet drainer-as-a-service kit |
| MITRE ATT&CK techniques | T1204 User Execution, T1213 Data from Information Repositories, T1528 Steal Application Access Token, T1566 Phishing, T1583 Acquire Infrastructure, T1585 Establish Accounts, T1587 Develop Capabilities, T1589 Gather Victim Identity Information, T1598 Phishing for Information, T1608 Stage Capabilities, T1657 Financial Theft, T1684.001 Impersonation |
| Motivation | FINANCIAL |
| Target sectors | retail investors, cryptocurrency, financial, individuals, decentralized finance |
| Target regions | United States, United Kingdom, Canada, Europe, Global |
| Detection rules | 9 (1 shown free) |
| Sources | 10 cited sources |
Detection coverage
As of 2026-06-17, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-0837 across Splunk SPL, Microsoft KQL and Sigma, covering 18 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.
Free detection: Unlimited ERC-20 approve() to Non-Allowlisted Spender — Exploitation
Detects the core approval-phishing exploitation step: an on-chain ERC-20 approve(address,uint256) call (selector 0x095ea7b3) whose amount slot is MaxUint256 (0xffff...ff), granting an unlimited spending allowance to a spender that is not on the organization's allowlist of known dApp/router contracts. This is the unlimited-allowance grant a victim is coached to sign on a fake high-yield investment dApp before funds are drained via transferFrom.
- Languages: Splunk SPL, Microsoft KQL, Sigma
- ATT&CK: T1566.002, T1204.001, T1528
- Data sources: evm:transaction, blockchain_transaction_monitor, mempool_feed
- Severity: high
- Confidence: high
Splunk SPL
index=blockchain sourcetype="evm:transaction" (input_data="0x095ea7b3*")
| eval selector=substr(input_data,1,10)
| eval amount_slot=substr(input_data,75,64)
| eval spender="0x".substr(input_data,35,40)
| where selector="0x095ea7b3" AND like(amount_slot,"ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff")
| search NOT [| inputlookup allowlisted_spenders.csv | fields spender]
| stats count min(_time) as first_seen max(_time) as last_seen values(to_contract) as token by from_address spender
| eval reason="Unlimited (MaxUint256) ERC-20 approval to non-allowlisted spender"
| table first_seen last_seen from_address spender token count reasonMicrosoft KQL
EvmTransactions_CL
| where TimeGenerated > ago(24h)
| where InputData_s startswith "0x095ea7b3"
| extend AmountSlot = substring(InputData_s, 74, 64)
| extend Spender = strcat("0x", substring(InputData_s, 34, 40))
| where AmountSlot == "ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff"
| where Spender !in ((AllowlistedSpenders_CL | project Spender_s))
| project TimeGenerated, FromAddress_s, Spender, ToContract_s, TxHash_s
| summarize Count = count(), FirstSeen = min(TimeGenerated) by FromAddress_s, Spender, ToContract_sSigma
title: Unlimited ERC-20 Approval (approve MaxUint256) to Unknown Spender
id: 7a1d2c9e-0837-4a01-bf10-approve001spl
status: experimental
description: Detects an ERC-20 approve(address,uint256) call granting MaxUint256 unlimited allowance to a non-allowlisted spender, the exploitation step of approval phishing.
references:
- https://www.chainalysis.com/blog/what-is-approval-phishing/
- https://attack.mitre.org/techniques/T1528/
author: Threadlinqs Team
date: 2026/06/17
tags:
- attack.initial_access
- attack.t1566.002
- attack.execution
- attack.t1204.001
- attack.credential_access
- attack.t1528
logsource:
product: blockchain
category: transaction
detection:
selection:
input_data|startswith: '0x095ea7b3'
input_data|endswith: 'ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff'
filter_known:
spender|in: 'allowlisted_spenders'
condition: selection and not filter_known
falsepositives:
- Legitimate one-time unlimited approval to a well-known DEX router or bridge (allowlist these spenders)
level: highKnown false positives:
- Legitimate unlimited approval to a well-known DEX router (e.g. Uniswap Universal Router) or bridge contract that is simply not yet on the allowlist
- Power users intentionally setting MaxUint256 to save gas on repeated trades
Tuning notes: Maintain allowlisted_spenders.csv with known router/bridge/aggregator contracts per chain. Tune by alerting only when the spender is a freshly-deployed contract (< 30 days) or an EOA, which is a much stronger phishing signal than an established router.
- Malicious EIP-712 Permit Signature Request from Untrusted dApp — Wallet Telemetry · unlock with TL Creds
- increaseAllowance / Repeated approve to Same Spender Across Wallets — Allowance Harvesting · unlock with TL Creds
- Lurk-Then-Drain: transferFrom Immediately After Fresh Inbound Deposit · unlock with TL Creds
- Wallet Connect + Approve Sequence from Newly-Registered Investment dApp Domain · unlock with TL Creds
- Permit2 Batched Signature Granting Multi-Token Allowance to Unknown Spender · unlock with TL Creds
- Multicall3 DELEGATECALL Atomic permit + transferFrom Drainer Bundle · unlock with TL Creds
- NFT setApprovalForAll to Unverified Operator — Post-Exploitation Persistence · unlock with TL Creds
- permit() Submission Followed by transferFrom Drain — Atomic Wallet-Drainer Bundle · unlock with TL Creds
Indicators of compromise
As of 2026-06-17, Threadlinqs Intelligence tracks 18 indicators of compromise for TL-2026-0837.
Technical analysis
Approval phishing is an on-chain social-engineering technique that abuses the legitimate token-authorization model of EVM (and TRON) smart-contract platforms rather than any software vulnerability. Instead of stealing a private key, the attacker convinces a victim to authorize a transaction or sign an off-chain message that grants the attacker's address an allowance over the victim's tokens. Once the allowance exists, the attacker can call transferFrom at any time to move funds out of the victim's wallet — instantly, or by lurking until the victim deposits fresh funds.
The abuse centers on three approval primitives. For ERC-20 tokens the attacker induces a call to approve(address,uint256) (selector 0x095ea7b3) or increaseAllowance, typically requesting the maximum uint256 value (0xffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff) so the allowance is effectively unlimited and persists until revoked. For ERC-721/ERC-1155 NFTs the attacker induces setApprovalForAll(address,bool) (selector 0xa22cb465), which authorizes the operator to move every token in a collection. The most dangerous variant abuses EIP-2612 permit (selector 0xd505accf) and Uniswap's Permit2 standard: the victim signs an off-chain EIP-712 message — never a visible on-chain transaction — and the attacker submits it on-chain to obtain the allowance, then drains in the same bundle. Observed drainer kits chain a permit and transferFrom atomically (e.g., DELEGATECALL through Multicall3) so a single signature grants unlimited USDC/USDT approval and immediately distributes stolen funds, with no protocol exploit or flash loan required.
The social-engineering wrapper is usually a pig-butchering / romance-investment scam: victims are coached off regulated exchanges into self-custody wallets, walked through 'connecting' to a fake high-yield investment or trading dApp by a 'mentor' figure using rehearsed lines and artificial urgency, and told the approval is a routine step to 'activate' trading. The lure surfaces as a bank red flag when customers with no crypto history suddenly wire large sums.
Law-enforcement and industry response has been substantial: Operation Spincaster (Chainalysis-led, launched July 2024) processed over 7,000 leads and addressed roughly $162M in losses across sprints in six countries, against an estimated $2.7B stolen via approval phishing between May 2021 and July 2024. Operation Atlantic (announced April 9, 2026; UK NCA with US Secret Service, Ontario Provincial Police, Ontario Securities Commission, Chainalysis and TRM Labs) identified more than 20,000 victim wallet addresses across 30+ countries, directly contacted over 3,000 victims, froze more than $12M in proceeds and traced a further $45M. Operation DeCloak addressed approval-phishing fund freezes/seizures in Delta, Canada.
MITRE ATT&CK techniques by tactic:
Execution
Collection
T1213 Data from Information Repositories
Credential Access
T1528 Steal Application Access Token
Initial Access
Resource Development
T1583 Acquire Infrastructure; T1585 Establish Accounts; T1587 Develop Capabilities; T1608 Stage Capabilities
Reconnaissance
T1589 Gather Victim Identity Information; T1598 Phishing for Information
Impact
stealth
Remediation
Immediate actions
- Review and revoke all unnecessary token allowances using an approval manager (e.g., revoke.cash or the wallet's built-in approval/permission manager); prioritize unlimited (MaxUint256) approvals and any setApprovalForAll operators
- Before signing, decode calldata: verify the spender address (second 32-byte slot, left-padded) against the intended dApp and confirm the amount is not 0xffff...ff
- Treat off-chain signature requests (permit / Permit2 / EIP-712) with the same scrutiny as transactions — verify spender, amount, token, and contract address
- If actively scammed, race the attacker by submitting an approve(spender,0) / revocation before the next transferFrom
Workarounds
- Set finite, per-transaction allowances instead of unlimited approvals where the dApp supports it
- Periodically audit approvals and revoke stale operators/spenders
Longer-term hardening
- Use a dedicated hot wallet with minimal balances for unknown dApp interactions; keep core funds in a separate cold/hardware wallet that never connects to dApps
- Enable wallet transaction-simulation / approval-warning features and clear-signing on hardware wallets
- Educate users that legitimate platforms never require unlimited approvals to an unknown address, and that 'mentors' coaching off regulated exchanges into self-custody are a scam signature
Affected and fixed versions
- Ethereum / EVM ecosystem — ERC-20 token allowance model (approve / increaseAllowance)
Vulnerable versions: ERC-20 approve(address,uint256); EIP-2612 permit; Uniswap Permit2 - Ethereum / EVM ecosystem — ERC-721 / ERC-1155 NFT operator approvals (setApprovalForAll)
Vulnerable versions: setApprovalForAll(address,bool) - Tether — USDT (ERC-20 and TRC-20)
Vulnerable versions: USDT-ERC20; USDT-TRC20
Timeline
- Start of Chainalysis tracking window for approval-phishing losses later estimated at ~$2.7B (through July 2024).
- During a Spincaster sprint, partners directly contacted a victim who revoked the malicious approval before a six-figure sum was drained.
- Chainalysis launches Operation Spincaster, a public-private initiative targeting approval-phishing and pig-butchering scams across six countries (US, UK, Canada, Spain, Netherlands, Australia).
- Tether, working with Chainalysis and the DOJ, froze approximately $225M in USDT held in perpetrators' wallets tied to a Southeast Asia pig-butchering compound.
- On-chain scams reach at least $14B (likely $17B) for 2025; average payment per scam address up 253% YoY; AI-augmented scams 4.5x more profitable.
- Law enforcement directly contacts more than 3,000 identified Operation Atlantic victims; one UK victim reported a loss exceeding £52,000.
- Operation Atlantic freezes more than $12M in suspected criminal proceeds and traces a further $45M in stolen crypto.
- Operation Atlantic announced: NCA/US Secret Service/Ontario Provincial Police/OSC/Chainalysis/TRM identify 20,000+ victim wallets across 30+ countries during a week-long action.
- Threat documented in Threadlinqs Intelligence Platform as TL-2026-0837.
- Chainalysis publishes explainer 'What Is Approval Phishing? How Scammers Drain Crypto Wallets'.
Sources cited
- Operation Atlantic disrupts more than $45 million in cryptocurrency fraud, freezes $12 million in stolen funds secretservice.gov · Advisory · 2026-04-09
- TRM Labs Supports Operation Atlantic: USD 12 Million Frozen and 20,000 Victims Identified trmlabs.com · Research · 2026-04-09
- Detecting Token Approval Phishing in Calldata ghostkit.net · Research · 2026-01-01
- USDC Permit Phishing Drain darknavy.org · Research · 2025-01-01
- Understanding Ethereum Token Approvals support.ledger.com · Vendor report · 2025-01-01
- What Is Approval Phishing? How Scammers Drain Crypto Wallets chainalysis.com · Blog · 2026-06-17
- Operation Spincaster: Disrupt & Prevent Losses in Crypto Scams chainalysis.com · Blog · 2024-07-01
- How Public-Private Collaboration Is Freezing Crypto Scam Proceeds (Operation Atlantic) chainalysis.com · Blog · 2026-04-09
- ERC-20 Authorization: How Permit and Permit2 Work, Risks, and Key Differences gate.com · Blog · 2025-01-01
- Avoiding Approval Phishing and Wallet Drainers: A Consumer Checklist coca.xyz · Blog · 2025-01-01
Previous and next in phishing
Drafted by the Threadlinqs research pipeline from 10 sources; published ; last updated —.