False-positive “high-risk cyber activity” flag blocks gpt-5.3-codex in Codex CLI
Summary:
My account is being incorrectly flagged for “potentially high-risk cyber activity,” and requests are being routed to gpt-5.2 instead of gpt-5.3-
codex. I’m a solo developer using Codex CLI for normal local development (Python app debugging, SQLite, specs/docs). No cybersecurity-related
activity intended.
Banner text (verbatim):
“⚠ 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”
Impact:
- Prevents access to the model I’m paying for (gpt-5.3-codex).
- Disrupts workflow and performance/behavior is different on fallback model.
When:
- First observed: Feb 18th, 04:00 UTC
- Still happening as of: Feb 18th, 05:00 UTC
Environment:
- Product: Codex CLI
- OS: [Windows + WSL2 Ubuntu / Linux / macOS] (if relevant)
- Workspace: local repo for a trading journal (Python/FastAPI/SQLite)
- Typical commands: rg, sqlite3, running Python scripts, gdb for crash traces
Request:
Please review and remove the flag / restore access to gpt-5.3-codex. If specific content triggered it, I’d like to know what pattern caused the
false positive so I can avoid it.
Attachments:
<img width="1342" height="60" alt="Image" src="https://github.com/user-attachments/assets/53865229-984c-47f6-ab08-d276b589b659" />
thread ID 019c3390-5f0c-7721-a288-1d0db306e14e
20 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
@slurrr, have you successfully gone through the Trusted Access verification process?
@etraut-openai Facing a similar issue. Just happened. In my case, I have submitted all verification documents, but I keep facing "Your identity couldn't be verified or your account is ineligible at this time". I don't do anything even remotely close to cybersecurity stuffs.
Having the same issue here.
@etraut-openai I have not
@slurrr, OK thanks for confirmation. That's the mechanism we have in place currently for restoring access to
gpt-5.3-codexif your account is flagged.As noted in our documentation: "We recognize that joining Trusted Access may not be a good fit for everyone, so we plan to move from account-level safety checks to request-level checks in most cases as we scale these mitigations and strengthen cyber resilience."
@etraut-openai I understand you all need to do something to mitigate, but shotgun blasting everyone (super small percentage, I know) to some biometric identification "Trusted Access" is overkill. The burden should lie on OpenAI, not the paying customer. You flagged a bunch of small time solo dev hobbyists from what it looks like. Hope you figure it out soon.
also having the same issue 019c6f7c-095e-78c1-844b-9733a2e513d0. I'm not doing anything cyber...
Same issue. Thread id: 019c6f83-321f-7df1-870b-b282baf4d832
This happened after I created a session to test the new sandbox. I'm a maintainer of a popular opensource project.
UPD
After upgrading it to version 0.104, the incorrect output was removed.
hitting me as well. went through verification twice, said succesful, but then shows "Your identity couldn't be verified or your account is ineligible at this time. If you think this is a mistake, please contact support." on /cyber. In the middle of my work day, this is not great at all.
<img width="1470" height="912" alt="Image" src="https://github.com/user-attachments/assets/f8706920-07fe-487d-bfcf-5ca968b73ddb" />
finally https://chatgpt.com/cyber shows me as verified, but the long wait was not an ideal customer experience at all and needs fixing.
@etraut-openai why do we even need to verified and give our legal information to OAI even tough we dont work with anything cyber related??? This is your safety mechanism issue, but the solutions you give here is for ALL of us to sign into Trusted access for Cyber or whatever that buzzwords mean.
hitting me as well. did verification twice, verification site said it was a success, but openai does not seem to recognize. unsubscribed from my account. just spent 199 EUR for nothing. this is unbearable. I will not return to open ai as a customer. this feels like fraud.
same here. however I have unsubscribed in the meantime. they better fix their process. I will not resubscribe again.
Happened to me as well.
Thanks for the bug report. We're continuing to tune our cyber safety classifiers to make them more accurate. You can read more about this here. If you see this again in the future and believe it's a false positive (i.e. you aren't doing anything related to cybersecurity), please use the
/feedbackslash command and choose "Safety Check" to report it. That will help us further tune our checks.Just got this too...
stream disconnected before completion: This user's access to gpt-5.3-codex-ab-control-1020-1p-codexswic-ev3 has been temporarily limited for potentially suspicious activity related to cybersecurity. Learn more about our safety mitigations: https://platform.openai.com/docs/guides/safety-checks/cybersecurity
I've experienced the same issue a few minutes ago, the ticket might have been closed, but this issue has not been fixed, as I had Codex remediating some CodeQL alerts, reviewing CodeQL alerts should not be flag as cyber security risk.
thread ID 019cb130-b161-73f0-b22b-e294f20c502a
I just got this working on a single page HTML with JS in VSCode using the Codex plugin. The same app I've been working on for days. All it does is allow me to track game progress and stored data in pocketbase. I asked AI to clean up some CSS and it threw this error below. Clearly this is some kind of bug they need to sort out. I'm on the free trial plan and it's been good so far.
I uploaded a feeback safety check request as instructed. 019c864f-b2eb-7b82-bf82-f82f18e7ad3e
I just fixing my app ui, LOL.
Ok somehow it worked again after i close my IDE and reopen the project.