OpenAI 修复 Codex 误删用户文件的安全漏洞

AI 摘要

OpenAI 修复了 Codex 中的一个安全漏洞，该漏洞曾导致模型在运行过程中未经用户许可删除真实文件。问题源于清理临时文件的命令误用了系统变量，指向了真实目录。OpenAI 已推出安全更新，要求 Codex 在执行删除前验证目标、创建新的临时文件夹，并加强了对高风险删除命令的检查。

正文节选

OpenAI fixes Codex bug that deleted real user files without permission OpenAI's Codex is no longer supposed to delete your data without asking. Several users had reported that GPT-5.6 Sol in Codex was autonomously wiping files while running. OpenAI has now shipped several safeguards in a security update. According to OpenAI, the root cause was a command meant to clean up temporary working files that could instead nuke actual user data. This happened when the model used system variables like $HOM

发布时间：2026-08-20 02:18

抓取时间：2026-08-20 02:39

来源机构：THE DECODER