Enterrpises need ability to enforce the --approval-mode setting at the enterprise-workspace level via the ChatGPT admin console.
Background:
- Developers currently have the option to configure Codex-1 with
--approval-mode full-auto, which automatically approves all AI recommendations. - At present, the ChatGPT admin panel does not offer a mechanism for workspace administrators to mandate or prohibit this mode for all users.
- There is no existing enterprise-level policy or flag to control the approval-mode behavior of Codex-1 agentic tasks.