Codex Trusted Access verification always fails / 'This content can't be shown' cybersecurity block

Q: Codex Trusted Access verification always fails / 'This content can't be shown' cybersecurity block | Vibe Coding Rescue

When Codex CLI/app keeps hitting 'This content can't be shown' and Trusted Access verification always fails, here's the real cause (a server-side false-positive classifier plus eligibility gating) and the realistic actions available today.

Codex false-flags ordinary work as a 'cybersecurity' risk and blocks it, while Trusted Access verification keeps failing as 'ineligible.' There's no confirmed fix — here's why, and what you can actually do now.

Symptoms

You ask Codex (CLI or desktop app) to do ordinary work, but instead of an answer you get blocked with:

> This content can't be shown. We take extra caution with cybersecurity requests. If you're a security professional, you may be able to apply for Trusted Access.

When you follow the link to the Trusted Access / verification flow, verification keeps failing:

> Could not start verification... may not be eligible for this verification flow

> Your identity could not be verified or your account is ineligible

It fires even on normal tasks — web/page maintenance, automation, devops, server management on your own systems, fixing vulnerabilities.

Cause

This is not a problem with your install or network. Run codex doctor and every check (app-server, auth, config, network, sandbox) comes back ok (GitHub #35256). The real trigger is OpenAI's server-side cybersecurity classifier — Trusted Access for Cyber, or 'TAC' — false-positively flagging legitimate work as a risky request.

And the reason verification 'always fails' is that Trusted Access is an eligibility-gated program. Ordinary Pro/Plus accounts get bounced out of the flow as 'ineligible' (#34236). In other words, the verification failing is expected behavior — you didn't do anything wrong.

No confirmed fix (being honest)

As of 2026-07-25 this is open with no OpenAI-confirmed fix or official workaround. The same symptom is piled up across many duplicate issues (#35256, #34236, #33130, #32225, #32095, #31294) with no maintainer resolution yet. What follows is not 'how to fix it' — it's what you can realistically do right now.

What you can do right now

Confirm it's not you — run codex doctor. If everything is ok, your install, login, and network are fine and the block is a server-side decision. Reinstalling or re-logging in won't clear it.
Report it precisely to OpenAI — copy the thread/trace ID shown with the block (it starts like 019f...) and file it on the openai/codex GitHub issues or in-product feedback, saying legitimate work is being false-flagged as cybersecurity. That trace ID lets OpenAI track your specific case — other users are already leaving trace IDs this way.
Only apply for Trusted Access if you actually qualify — real security professionals/orgs can apply via the form, but individual Pro/Plus accounts usually fail with 'ineligible.' Repeated failure is expected and not your fault, so don't burn time re-trying it in a loop.
Narrow the request and retry (not guaranteed) — instead of one big 'run a full security scan / audit all vulnerabilities,' split it into one file / one function, drop attack-flavored words like 'exploit/attack/payload,' and phrase it plainly ('refactor this code,' 'fix this bug'). This may trigger the classifier less often, but false positives are reported even on purely local work, so there's no guarantee it gets through.
Fall back if it's urgent — if the task has to ship now, the practical move is to hand that one task to another coding agent that supports auto-fallback / a different model.

Note

The block message can appear while background subagent work keeps running (#33130), so the UI may look stalled while execution — and billing — continues. If a long task gets blocked mid-run, stop it and re-issue it in smaller, plainer pieces as in step 4.