Recently, a user in the Linux community reported that while using the ClaudeOpus4.5 version of the AI tool Antigravity to execute tasks, the tool automatically ran a deletion command. As a result, it not only cleared the current working folder but also mistakenly deleted all contents in the parent directory, causing the loss of multiple important project files. This incident has drawn widespread attention from developers regarding the safety of AI-assisted tools. Experts analyzing the situation point out that such operational errors expose potential risks of AI tools in command-line automation. They recommend that users strictly configure security mechanisms, regularly back up data, and enhance testing when using these tools. This case serves as a warning to AI developers to optimize algorithms and add operation verification steps to prevent similar data loss incidents from happening again, providing valuable security lessons for AI users.
