I was utterly stunned after seeing this. In my understanding, engineers should have a strong sense of cybersecurity, especially engineers at an exchange. They should always check the hash value before installing any software and double-check that it's from an official source. But the first engineer, in order to use an AI tool, directly downloaded a tampered Homebrew repository. The second engineer went even further, downloading the interview questions to his work computer and running them. This is unbelievable in itself. Then the Wallet DevOps' permissions were far too high; a single token could trigger the Wallet signing service. And how could internal AI tools be used to lateral movement and control other engineers? After watching it, I really don't think it was a problem of AI triggering any zero-day vulnerabilities; it was just a bunch of outrageous errors strung together.