Single attacker hits Fetch.ai and NuNet, nearly $2 million lost

CoinMarketCap reported that blockchain security firms Blockaid and PeckShield said the same attacker struck two AI-related crypto projects—Fetch.ai and NuNet—on Sept. 19, with combined losses of about $2 million at the time. The incidents were followed by sharp declines in both tokens. Blockaid said the attacker siphoned roughly 8.7 million FET—about $1.53 million—from a Fetch.ai token swap contract. Researchers added that the receiving wallet later surfaced in the NuNet exploit, a key datapoint tying the two events to a single actor. Blockaid also published the attacker-linked wallet addresses and an example transaction to help projects and exchanges flag and potentially block related fund flows. In the NuNet case, the attacker used NuNet's deployer account to mint about 408.5 million unauthorized NTX tokens, valued at roughly $463,000 based on the report's cited price. Investigators said the newly minted NTX was sent to the same wallet that received the stolen FET, reinforcing the link between the two attacks. Security researchers said the pattern points to weaknesses in contract permissioning or privileged key management. When deployer permissions or upgrade paths are abused, attackers can drain existing assets and also create new token supply. After the attacks, the attacker rapidly swapped a substantial portion of the proceeds into 546 ETH, worth about $1.44 million, a common post-exploit tactic aimed at complicating tracking and potential freezes. On the news, NuNet's NTX fell about 65% to 70%, while Fetch.ai's FET dropped roughly 10%. As of publication, neither Fetch.ai nor NuNet had issued an official statement. Key figures: FET transferred—about 8.7 million; unauthorized NTX minted—about 408.5 million; attacker received—546 ETH. The report added that security firms view the incident as highlighting a recurring risk for AI and computing-focused crypto projects: if a deployer account or token conversion contract is compromised, losses can cascade across multiple projects.