OpenAI is rebuilding Codex's plumbing: persistent cloud environments, a voice-enabled CLI, sharper code review, and a dedicated security scanner.
OpenAI says Codex now supports reusable cloud development environments, letting a coding session started on one device pick up on another without rebuilding the setup each time. The company also revamped Codex's command-line interface to accept voice commands, added new code review tools, and launched a separate security product. That security tool scans repositories for vulnerabilities and drafts fixes for whatever it finds.
The persistent environments matter more than the voice-control headline: they push Codex toward being infrastructure you keep, not a session you rebuild every time you open a terminal. Bundling an automated vulnerability scanner into the same product puts OpenAI in the same lane as dedicated application-security vendors, not just rival AI coding assistants.
Whether "prepares fixes" means patches you can trust or plausible-sounding suggestions is the detail worth testing before anyone lets Codex near a production repo.