Web3 Security Under Siege: AI-Powered Scams and Exploits Lead to Over $45M Losses in October

Web3 security faced unprecedented challenges in October 2025, with artificial intelligence and automated exploits driving a wave of attacks that resulted in combined losses exceeding $45.8 million. According to a report by GoPlus Security, the month witnessed 16 high-profile incidents, including hacks, phishing scams, and social engineering attacks, highlighting a critical vulnerability in the ecosystem.

One of the most significant breaches targeted SBI Crypto, a cryptocurrency exchange, which suffered a $21 million loss in assets including Bitcoin (BTC), Ethereum (ETH), Litecoin (LTC), Dogecoin (DOGE), and Bitcoin Cash (BCH). Investigations by ZachXBT and Cyvers revealed laundering patterns through Tornado Cash that resembled those of North Korean hacker groups like the Lazarus Group, though SBI Crypto did not confirm the source.

Phishing scams escalated with the rise of Phishing-as-a-Service platforms, which use AI to clone legitimate sites and launch large-scale campaigns. In the GMGN incident, 107 users approved fake transactions on a cloned site, leading to losses over $700,000. Separate cases involved users losing $325,000 and $440,000 after signing malicious commands like 'increaseAllowance' and 'permit'.

Other major exploits included the Astra Nova hack, where a compromised official account triggered a sell-off of its native token RVV, costing $10.3 million, and the Garden Finance vulnerability, which resulted in a $10.8 million loss. Additionally, honeypot tokens—malicious contracts that block users from selling or withdrawing funds—surged by 600% month over month, with 2,189 cases detected. Most were on BNB Chain (1,780), followed by Ethereum (216) and Base (131).

GoPlus Security emphasized that attackers are leveraging AI and automation to reduce costs and scale threats, expanding the attack surface beyond code vulnerabilities to exploit user trust directly.