A recent analysis by @AdiFlips suggests that the code used in the Balancer hack may have been generated with AI assistance. The hacker left console.log debugging information in the on-chain code, which is unusual for professional hackers who typically remove such logs. The presence of nonsensical content like "Done with amts1" in the logs indicates the potential use of AI-generated templates, possibly from large language models (LLMs), in crafting the attack code.