I hope someone from the OpenAI team can help resolve this issue, as Support has said that it is outside their control and that they cannot override the verification result. However, the email I received specifically instructed me to contact OpenAI Support if I believed there was an error.
I believe there may be a bug in the Trusted Access for Cyber verification flow at chatgpt.com/cyber.
I was previously verified and approved for Trusted Access for Cyber. On July 7, 2026, I received an email stating that my access had been removed only because my account was no longer on an eligible paid plan. The email explained that I could restore an eligible paid plan and request access again.
I have now subscribed to the Pro plan and attempted the verification process again several times. However, my request is neither approved nor clearly denied. The page simply asks me to retry verification and shows the following message:
“Your identity couldn’t be verified or your account is ineligible at this time. If you think this is a mistake, please contact support.”
I have tried the verification process multiple times using the same identity documents that were previously accepted, but I continue to receive the same result.
This appears to be a possible account-state or verification-flow bug because:
I was already approved previously.
My access was removed specifically because I was no longer on a paid plan.
I restored an eligible paid plan as instructed.
The verification flow does not provide a final approval or denial and only keeps asking me to retry.
I contacted OpenAI Support, but they said they could not provide further details or override the result. I am not asking for a manual override. I am asking for the verification flow or my account status to be reviewed in case the previous removal of Trusted Access left my account in an incorrect state.
Has anyone else experienced this issue after restoring an eligible paid plan? I would appreciate it if someone from the OpenAI team could investigate and help resolve it.
I couldn’t find a solution, and OpenAI Support wasn’t able to resolve the issue either. I ended up canceling my ChatGPT subscription and have switched to Claude, since my Claude CVP has been approved.
Has anyone got access to TAC in the last two months? I am almost certain its broken for literally everyone but it seems to be a non-issue for OpenAI as other posters claimed its been 2+ months of the same issue.
This is very concerning to me since defenders are essentially unable to secure their code.
I tried to verify for the first time ever, persona accepted my docs and verified my identity but it still says “Your identity couldn’t be verified or your account is ineligible at this time. If you think this is a mistake, please contact support.”, don’t know if anyone else had that issue as well
Yes, it’s the same bug that’s been raised by multiple users for multiple months. OpenAI is aware that the bug exists but there has been no public acknowledgement or communication regarding the bug.
I was on a 20X Pro plan, but TAC can’t even verify it and keeps giving me an error telling me to retry verification, even though I was previously verified on TAC with the same government ID. After contacting support and trying multiple times, I gave up and moved to Claude. It would be great if you could contact the engineering team and fix this TAC issue for all of us.
I did the same. Support has kept me waiting three weeks with nothing..this is really concerning because we literally can’t defend our code from bugs that even GPT writes
It’s been over a month for me..others have been unable to do any cyber security work, including bug fixes for three or more months. Very concerning that cyber security is not a concern for OpenAI.
Seems I’m running into this issue as well, for which I’ve submitted a support case (Case Number: 12688082). I have a legitimate use case and would like to use Codex with Cyber guardrails lifted to work through and pass the OSAI exam, however due to this “bug” it looks like I’m going to be forced to switch to Anthropic. Support for this seems to be non-existent.
Excellent business model you have here OpenAI, keep this up and I’m sure you’ll stay in business at least for the next few months.
