According to Wu, @AdiFlips analyzed that the Balancer hacker left console.log debugging information in the on-chain code, suggesting that the attack code may have been generated with the assistance of AI (written by LLM). He pointed out that hackers usually do not leave debugging logs in official code, and such mistakes are often copied from templates generated by LLM, with meaningless content like "Done with amts1" appearing in the logs, further indicating that the attack code may bear obvious traces of AI coding.