Security advisories
This page provides security advisories for OpenAI products to help users understand and address potential security risks. Advisories include information about affected products and versions, the potential impact of each vulnerability, and available security updates or mitigations.
8 results
- CVE-2026-110930Opening a project from a Windows library could run code in Codex DesktopLearn more
- CVE-2026-11326A script on an OpenAI subdomain could read browser history and control tabs in AtlasLearn more
- CVE-2026-14898Remote Markdown images could expose session data in Codex Desktop for macOSLearn more
- CVE-2026-19590Git hooks could run code without approval in Codex DesktopLearn more
- CVE-2026-19591A Git command could run through PowerShell without approval in Codex CLI and Codex DesktopLearn more
- CVE-2026-19592Git could run a filesystem monitor without approval in Codex CLI and Codex DesktopLearn more
- CVE-2026-19593Git attribute filters could run code when a user opened a workspace in Codex DesktopLearn more
- CVE-2025-59532A command could change the sandbox's writable directory in Codex CLI and the Codex IDE extensionLearn more