Flagged while already being verified

Same issue for me. Verified for a few weeks and now get "Your account was flagged for potentially high-risk cyber activity. Requests may be slower while additional verification is applied. To regain faster access, apply for trusted access: https://chatgpt.com/cyber or learn more: https://developers.openai.com/codex/concepts/cyber-safety" with every request.
And yes, the requests are indeed slower now.

Same - happened after gpt-5.5 release

same also got my account revoked, doing the exact same testing/activity as i was before with gpt 5.4 all week last week up until release.

Same, doing malware analysis, continuing the same thread that had been working with 5.4

same, i'm verified and simply continued with my 5.4 workflows and got flagged immediately

Same thing is happening to me. Setting aside the fact that it's flagging accounts that are already cyber verified, which needs to be looked into on its own, the flagging should at least be scoped to individual threads, not entire accounts. Just end the flagged thread, similar to how Claude handles it, or throttle that specific thread. It shouldn't degrade performance across other threads that for example are running normal frontend work..

Some information from my side:
Uploaded thread: 019dbeb7-1a36-7bc2-ba7d-360835fd675f
Codex CLI version: codex-cli 0.124.0
Platform: macOS 26.4.1
Model: gpt-5.5
Subscription: 200$ Pro

Trusted Access verification: verified since this new cyber security process was introduced 2 months ago.

I used GPT 5.5 in the exact same way I did 5.4. Never happened before.
There's no way to reverify which should also not be needed, as I'm not using it in a malicious way at all.
Responses now take a long time since I got this message.

Same here. It’s honestly hilarious that I was just messing around with a CTF and somehow got my verified Pro account nerfed forever. I did the exact same thing with 5.4 and had zero issues.

Same here - doing fuzzing research and now I also received this message although I have already verified my account before...

Very frustrating, because if I knew this wasn't allowed (even after being PRO account with TAC verified), I would just not do it.

OpenAI - Access Deactivated [C-NZpXX7gmcPVp]

@etraut-openai how can we tell if this is a false warning or if we're actually being slowed?

I have the same problem. My country has censorship issues. I asked for help setting up a VPN. Today, my account was blocked. I paid for a subscription for 200+ euros, about 3-4 days ago, and the money was stolen.
<img width="545" height="270" alt="Image" src="https://github.com/user-attachments/assets/6023ae69-b105-421f-9003-3cf5fdc86a22" />

I also got flagged, though I'm already cyber-verified. Was doing CTF challenges with GPT 5.5.

After essentially not interacting with any of the chats that could have gotten me flagged since I opened the ticket, the message disappeared. Around 12 hours ago, it still displayed the following:

Your conversations have multiple flags for possible cybersecurity risk. Responses may take longer because extra safety checks are on. To get authorized for security work, join the Trusted Access for Cyber program: https://chatgpt.com/cyber

Now the question is whether the flags have simply expired or whether the issue has been resolved. Since it appears that others have had their entire accounts terminated, I would appreciate an update, as I would not want to risk that. TAC does not appear to have been applied correctly.

I am also wondering whether I can now safely continue the thread I uploaded without having to worry.

@Acurisu I finally had my account reinstated (although I got downgraded to free and had to re submit for pro even though my pro sub got reupped 4 days ago). Also received an email stating my account was revoked in error. I've been able to continue where I left off with no errors or warnings as of last night.

@cruzanstx Thanks for the update. That is very reassuring to hear, except for losing Pro. I nonetheless hope for some official closure on this matter because it’s unclear when someone could get flagged next. Did you have to fill out the appeal form to get your account back?

Same issue.
I even tried applying for the verification but for some reason it keeps failing.

<img width="855" height="600" alt="Image" src="https://github.com/user-attachments/assets/ab6d5ca1-3833-4b85-a592-a0b548e276b5" />

@etraut-openai Many of us would appreciate some official clarification regarding this issue. It seems that most of the people affected have already been cyber-verified, and were flagged even while operating in accordance with allowed use guidelines. Since the action taken has been severe in some cases, including account suspensions made in error, affecting substantial paid plans, we would appreciate some clarity on the issue, and whether it has been addressed adequately, and what to do in the event of a false positive. As far as I have seen, there has been little official communication about this issue relative to how disruptive it has been. Thank you.

This no longer appears as an issue for me - I am running v0.125.0 and stopped getting the warning yesterday.

I am experiencing the exact same issue with Codex 5.5, even though I have successfully completed the verification chatgpt.com/cyber for Trusted Access.

My account was suddenly suspended by an automated system flag. My development workflow involves building a benign marketing intelligence product that tracks and analyzes company mentions in media. This requires running localized web scraping scripts and QA tools alongside my Codex IDE environment to test coverage.

It seems the current heuristic classifier for GPT-5.5 incorrectly flags web monitoring automation as high-risk cyber activity, completely disregarding the active Trusted Access for Cyber (TAC) status. I have submitted an official appeal form, but this is a serious false-positive bug affecting legitimate developers.

<img width="559" height="121" alt="Image" src="https://github.com/user-attachments/assets/3724c070-8eaa-4d87-8b39-1ea4f39338f8" />

Experiencing the same issue — Pro subscriber, all conversations flagged as cybersecurity risk, and the Trusted Access verification flow itself is blocked ("We couldn't start verification"). No security-related work being done, just regular server maintenance and web development. Filed #22988 with comprehensive details. This appears to be a systemic problem with the classifier's false-positive rate.

Adding a current data point from the same failure pattern. Our related report is #22988.

As of 2026-05-17 Asia/Seoul, the account is still being flagged during normal paid development/admin work, and the Trusted Access flow still cannot be started. OpenAI Help Center case 08901525 says support cannot manually remove the warning or directly enable Trusted Access eligibility.

This leaves affected paid users in a closed loop: regular work is degraded, the advertised verification path is blocked, and support cannot unblock it. OpenAI should provide an official remediation path and consider API credits or equivalent service credit for users whose paid work has been materially blocked or degraded by these false positives.