Skip to main content
• 6 min read

JaredFromSubway MEV Bot Drain:
$7.5M Lost to Counter-MEV Approval Honeypot

$7.5M was drained from the jaredfromsubway.eth MEV bot after fake token routes left ERC-20 approvals open and attacker contracts swept WETH, USDC, and USDT.

AUTOSEC.DEVAUTOSEC.DEV
JaredFromSubway MEV Bot Drain: $7.5M Lost to Counter-MEV Approval Honeypot
  • Incident Date: June 20, 2026
  • Target: jaredfromsubway.eth MEV bot on Ethereum
  • Target Overview: jaredfromsubway.eth is a highly active Ethereum MEV and sandwich-trading bot. The incident affected the bot's own treasury and approval surface, not a reported DEX protocol vault, bridge, or user wallet compromise.
  • Total Loss: Approximately $7.5 million in confirmed on-chain reporting; the bot operator reportedly claimed a higher $15 million loss, but that discrepancy has not been publicly reconciled.
  • Reported Stolen Assets: 1,474.58 WETH, 2.87 million USDC, and 2 million USDT.
  • Reported Fund Flow: Stolen assets were swapped into approximately 4,400 ETH, with 1,000 ETH deposited into Tornado Cash.
  • Attack Vector: Approval abuse / counter-MEV honeypot / automated decision-logic failure

Incident Review & Technical Details​

1. Attack Path​

  1. The attacker targeted automated MEV behavior: Blockaid reported that the exploit involved the jaredfromsubway.eth MEV bot on Ethereum. Instead of exploiting a conventional smart-contract bug, the attacker targeted the bot's automated execution logic and the approvals it generated while chasing apparent trading opportunities.
  2. Fake routes were prepared as bait: TechTimes, citing Blockaid's analysis, reported that the attacker deployed 66 fake token contracts and fake liquidity routes designed to look like profitable MEV paths. The trap reportedly imitated familiar assets such as WETH, USDC, and USDT closely enough to lure the bot's automated pattern-recognition logic.
  3. The bot granted approvals to attacker-controlled contracts: Blockaid said attacker-controlled contracts tricked the MEV execution system into granting token approvals. Those approvals were later used to drain funds, meaning the bot's own permission grants became the theft mechanism.
  4. A single transaction swept the exposed balances: Blockaid linked the exploit transaction 0x2be8704f5a59b69e0b71f64aefdb99eb0e8ae9fb3926147c581910d71bcf3e65, which Etherscan records as successful on June 20, 2026 at 18:49:11 UTC. Public reporting placed the swept assets at 1,474.58 WETH, 2.87 million USDC, and 2 million USDT.
  5. Funds were consolidated and moved into Tornado Cash: PeckShieldAlert reported that the attacker swapped the stolen assets into approximately 4,400 ETH and had already deposited 1,000 ETH into Tornado Cash.

2. Impact Scope​

  • Protocol-Level Loss: The canonical confirmed loss figure for this post is approximately $7.5 million, based on PeckShieldAlert, Blockaid, Blockchain.news, and TechTimes reporting.
  • Affected Component: The affected component was the jaredfromsubway.eth bot's execution and approval workflow, not a reported exploit of Ethereum itself, a DEX core contract, or a public token contract vulnerability.
  • Permission Surface: The incident shows how persistent ERC-20 approvals can become a compounding risk surface for automated trading systems that interact with many token contracts at high speed.
  • Behavioral Risk: The exploit did not require a human to sign a phishing transaction. The bot's automated decision logic allegedly approved attacker-controlled contracts during what appeared to be normal MEV execution.
  • Loss Discrepancy: TechTimes and Mallory noted that the bot operator claimed a higher $15 million loss and offered a bounty, while the reviewed on-chain security reporting identified approximately $7.5 million in confirmed drained assets.
  • Fund-Flow Risk: The conversion into 4,400 ETH and the 1,000 ETH Tornado Cash deposit reduce near-term recovery visibility.

3. Official Statements​

  • Blockaid: Blockaid said its Exploit Detection system detected an Ethereum exploit involving the jaredfromsubway.eth MEV bot. It attributed the incident to attacker-controlled contracts tricking an automated MEV execution system into granting token approvals that were later used to drain funds.
  • PeckShieldAlert: PeckShieldAlert reported that Specter had identified the drain, listed the stolen asset mix, and traced the swap into approximately 4,400 ETH plus the 1,000 ETH Tornado Cash deposit.
  • Blockchain.news: Blockchain.news summarized the incident as a $7.5 million drain from the JaredFromSubway MEV bot, with stolen WETH, USDC, and USDT converted into ETH and partially routed through Tornado Cash.
  • TechTimes: TechTimes described the incident as a counter-MEV honeypot in which fake token contracts and approval persistence turned the bot's own ERC-20 permissions into a drain mechanism.

4. Investigation Progress​

The exploit transaction and the core approval-abuse mechanism are public, but the reviewed sources did not identify a final technical postmortem from the bot operator, a full attacker address cluster, a confirmed recovery, or an explanation for the gap between the approximately $7.5 million on-chain reporting and the operator's reported $15 million claim.

Recommended response steps for automated DeFi and MEV systems:

  • Inventory all active ERC-20 approvals held by the bot, including approvals created during failed, reverted, or partially executed routes.
  • Enforce approval TTLs, per-route spend caps, and automatic revocation after each execution path finishes.
  • Treat unknown token contracts, clone assets, fake stablecoins, and attacker-seeded liquidity pools as adversarial inputs, not just low-liquidity opportunities.
  • Add simulations that run candidate routes against malicious token behavior, approval persistence, transferFrom() replay, and delayed drain scenarios.
  • Separate treasury balances from hot execution balances so a route-level approval failure cannot expose the full operating wallet.
  • Publish the full affected address list, approval list, exploit transaction sequence, attacker-controlled contracts, and recovery or bounty status.

AUTOSEC.DEV Solution​

The JaredFromSubway drain shows that automated DeFi systems need security testing for decision logic and approval lifecycle risk, not only source-code correctness.

  1. Secure Code Review - The reported failure path centered on ERC-20 approvals granted by an automated MEV execution system to attacker-controlled contracts. AUTOSEC.DEV reviews approval lifecycles, allowance scope, route validation, hot-wallet separation, and post-trade revocation logic so automated trading systems do not accumulate silent transferFrom() exposure.
  2. Penetration Testing - The attacker allegedly spent weeks preparing fake token contracts and routes that looked profitable to the bot. We reproduce adversarial route construction on forks, including counterfeit assets, malicious helper contracts, unconsumed approvals, and delayed drain attempts, so teams can test how automation behaves under hostile market inputs.
  3. Incident Response - Because proceeds were swapped into approximately 4,400 ETH and 1,000 ETH was routed into Tornado Cash, response depends on fast evidence preservation. AUTOSEC.DEV supports approval-surface reconstruction, transaction replay, attacker contract clustering, exchange or bridge coordination, disclosure support, and remediation validation.

Reference​