Attack on MEV Giant: Jaredfromsubway.eth Bot Lost Over $7.5 Million

The well-known MEV bot Jaredfromsubway.eth, which long dominated the field of sandwich attacks on Ethereum, fell victim to its own automation. As a result of a targeted attack, malicious actors managed to withdraw assets worth over $7.5 million. This incident is not just another hack, but a serious signal that even the most sophisticated algorithms can be deceived.
Analysis of the incident shows that the attack was unconventional. The attacker did not directly hack the bot's contracts or use classic phishing schemes. Instead, an entire network of dozens of fake token contracts was deployed, disguised as popular assets: WETH, USDC, and USDT. These tokens were linked to fake liquidity pools that mimicked profitable trading opportunities.
The MEV bot, programmed to seek arbitrage and sandwich attacks, "took the bait." The automated trade execution system granted the attacker's auxiliary contracts permission to spend real funds. After that, the attacker only needed to activate all the embedded backdoors with a single transaction and withdraw the assets. Part of the stolen funds has already been sent to the Tornado Cash mixer, as confirmed by Arkham data.
Context: The Scale of the Threat
It is worth noting that Jaredfromsubway.eth was not just one of the MEV bots, but a true giant. Estimates suggest that annual trader losses on Ethereum from sandwich attacks amount to about $60 million. Moreover, from November 2024 to October 2025, approximately 70% of all such operations on the network (60,000–90,000 per month) were linked to this bot. In June 2024, it even became the largest consumer of gas on the Ethereum network.
Expert opinion: This incident is a vivid example of how complexity and automation become the Achilles' heel of DeFi. Attackers no longer hack code; they deceive logic. For the community, this is a lesson: blindly trusting MEV bots means risking everything. The market must move toward more transparent and verifiable trade execution mechanisms.