Codex App: TAC(Trusted Access for Cyber) cannot be started, cybersecurity-risk block is sticky

What version of the Codex App are you using (From “About Codex” dialog)?

Version 26.715.31925

What subscription do you have?

ChatGPT Pro (individual)

What platform is your computer?

Microsoft Windows NT 10.0.26200.0 x64

What issue are you seeing?

Even after normal usage, the app repeatedly shows cybersecurity-risk blocking prompts.
Attempting to follow the remediation path at https://chatgpt.com/cyber fails with error messages instead of allowing verification.

Full message examples (PII redacted):

• "Your conversations have multiple flags for possible cybersecurity risk."
• "Could not start verification. You may not be eligible for this verification flow right now."
• "Your identity could not be verified or your account is ineligible at this time."

This blocks normal workflow even for standard, non-offensive tasks (web/page maintenance, automation, coding, devops, server management on owned systems).

What steps can reproduce the bug?

1. Sign in with a ChatGPT Pro account in Codex (desktop/app).
2. Start normal development/maintenance conversation.
3. Observe repeated cybersecurity-risk warnings and/or blocked responses.
4. Open the Trusted Access for Cyber page link (https://chatgpt.com/cyber).
5. Click through and attempt verification.
6. Verification fails with one of:

• "Could not start verification. You may not be eligible for this verification flow right now."
• "Your identity could not be verified or your account is ineligible at this time."

1. Repeat conversations remain blocked/flagged.

Session id / work id: 019f7c4c-144f-75f3-890c-bbe4b07fc979
Token/context details: Issue appears immediately; not related to token limit exhaustion.

What is the expected behavior?

Normal usage should continue with only expected safety checks, or
TAC verification flow should be available and complete successfully for eligible accounts.

Additional information

This appears to be a false-positive/eligibility dead-end loop.
Existing remediation path is effectively unreachable.
No new code execution or exploit behavior is involved; this is a workspace availability/security-classifier flow issue.
I can provide screenshots or additional IDs if needed.