Share a picture from @ 80vul for a general understanding of yesterday's classic AI agent incident where it was injected with prompt words, resulting in stolen coins. However, here are a few details to supplement and optimize: 1. The so-called Grok wallet (stolen wallet) is actually unrelated to the official Grok, and can be considered to be essentially a wallet generated by @ bankrbot for X @ grok, with private key permissions in the third-party wallet service that Bankr relies on. BaseScan also corrected the marking (Grok ->Bankr 1) https://basescan.org/address/0xb1058c959987e3513600eb5b4fd82aeee2a0e4f9 2. The Bankr Club member whose wallet was stolen was indeed activated, but it was not directly activated with an NFT. It should be a centralized mechanism, so there is no evidence on the chain for this part. However, there is confirmation of activation from Bankr: https://(x.com)/bankrbot/status/2051005172202358526 3. The highlight of the prompt word injection is to use the power of @ grok to hit @ bankrbot, which was stolen https://basescan.org/tx/0x6fc7eb7da9379383efda4253e4f599bbc3a99afed0468eabfe18484ec525739a The case is classic, so the details need to be treated rigorously. As for whether it is a script or not, it cannot be completely denied. Grok was indeed taken advantage of and labeled as a thief by too many people, but the protagonist of the incident was not Grok