GPT-5.3 requests automatically routed to GPT-5.2 under “Trusted Access for Cyber” (Pro subscription)

What issue are you seeing?

As a ChatGPT Pro subscriber, my requests to GPT-5.3 Codex are being automatically routed to GPT-5.2.

The UI shows:
“Your request was routed to GPT-5.2.”

The reason displayed is related to “Trusted Access for Cyber”, stating that my identity could not be verified or that my account is ineligible.

However, the prompts submitted are standard software engineering tasks (e.g., frontend updates, file uploads, database structuring, i18n handling) and are not related to cybersecurity, offensive security, exploitation, or vulnerability research.

This routing happens consistently across normal development prompts.

Example prompt (redacted):
“Inspect the repo structure and propose a plan to upload a multilingual PDF to Convex File Storage and update the frontend to resolve the correct file per locale.”

No security-related keywords were used.

What steps can reproduce the bug?

1. Open Codex app.
2. Select GPT-5.3.
3. Submit a normal software development prompt (non-cyber related).
4. Observe that the request is automatically routed to GPT-5.2.
5. UI indicates routing due to “Trusted Access for Cyber”.

This happens repeatedly across different normal development prompts.

What is the expected behavior?

As a Pro subscriber, GPT-5.3 should handle standard software engineering prompts without being automatically downgraded.

If “Trusted Access for Cyber” applies only to specific categories of requests, then:

• Non-cyber prompts should not trigger routing.
• The classification criteria should not affect general development usage.

Expected:
GPT-5.3 processes normal engineering prompts without fallback to GPT-5.2.

Additional information

• I am located in the EU.
• I have not submitted any offensive security or vulnerability exploitation prompts.
• This appears to be either:

a) Misclassification of normal dev prompts as cyber-related, or
b) An account-level restriction incorrectly applied.

Requesting clarification on:

• Why my account is flagged
• Whether GPT-5.3 access is conditionally gated
• Whether this is intended behavior or a bug