Flagged for high-risk cyber activity when not even using gpt-5.3-codex

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

• #12064
• #12060

Powered by Codex Action

<img width="753" height="605" alt="Image" src="https://github.com/user-attachments/assets/c4e8d3bd-d458-414f-9b6a-1ff918a80e01" />

did you fix the issue ? can you tell me how to fix it ,thanks a lot bro

Same here, and even though I'm already ID verified for the OpenAI API, apparently now I "can't be verified" with drivers license or passport. Mind boggling:

<img width="645" height="251" alt="Image" src="https://github.com/user-attachments/assets/669d9cdb-8fd8-4d4f-a999-068c8d53bd21" />

did you fix the issue ? can you tell me how to fix it ,thanks a lot bro

I reverted back to codex 0.101.0. I don't get the warning every time I send a message like I was in codex 0.103.0.

@rocky-m, if you're not trying to use gpt-5.3-codex, you can ignore the error message. The fact that you're seeing it when you're not using gpt-5.3-codex is a bug. We're working on a fix for that.

> did you fix the issue ? can you tell me how to fix it ,thanks a lot bro I reverted back to codex 0.101.0. I don't get the warning every time I send a message like I was in codex 0.103.0.

it seems work , hope that is the true fix , thanks

same man, in what gods name did i do that is a cyber threat?? Fix your shitty screening system please

same issue here i updated to 0.103.0 and i am getting flagged for security issues - cannot finish my work now :-(

019c6e1c-e4ca-73f1-985d-4c9d9e98f09e

Confirmed on Windows 10 LTSC after upgrading to codex-cli 0.103.

My repro is a bit different from the common “5.3 → 5.2” reports: I get the cyber-safety warning while already running gpt-5.2 (High).

Environment

• Windows 10 LTSC (native PowerShell)
• codex-cli 0.103
• Two terminal windows, same repo/workspace
• Same benign prompt in both windows (minimal localized fix in SetupComplete.cmd only)

Result

• Window A (gpt-5.3-codex High): runs fine, completes normally (never saw the warning here).
• Window B (gpt-5.2 High, explicitly selected): consistently and repeatedly prints:

⚠ Your account was flagged for potentially high-risk cyber activity and this request was routed to gpt-5.2 as a fallback. To regain access to gpt-5.3-codex, apply for trusted access:
  https://chatgpt.com/cyber or learn more: https://developers.openai.com/codex/concepts/cyber-safety

This is nonsensical because the selected model is already gpt-5.2 (a “self-fallback” message). Looks like a routing / model-id mismatch or a false-positive warning condition, not an actual account flag (since gpt-5.3-codex works in parallel).

I also submitted in-product feedback from the affected Codex session: thread ID 019c6e90-7cf9-7130-a632-dacd2dedaa37.

If helpful I can paste the exact prompt and my config.toml excerpt (notably:
[notice.model_migrations] "gpt-5.2" = "gpt-5.3-codex").

You should see the warning only when running gpt-5.3-codex. If you're using a different model such as gpt-5.2 and seeing this error, that's a bug, and you can just ignore the error message. We're working on addressing this bug, and we should have a fix out soon.

Updated to v0.104.0 and in the scenario I originally described, I don't get the warning anymore.