False positive policy flag on harmless prompts, automatic model downgrade #46753

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

Latest desktop Codex App

What subscription do you have?

pro x5

What platform is your computer?

Windows 11, x64

What issue are you seeing?

When I send normal, harmless technical questions (e.g. about measuring current of a 220-ohm resistor with a multimeter), I keep getting this error:
Invalid prompt: your prompt was flagged as potentially violating our usage policy. Please try again with a different prompt

This is a false positive. The prompt contains no prohibited content.
Meanwhile the model automatically downgrades itself continuously:
GPT-6 Astra → GPT-5.6 Sol → GPT-5.5o.
I cannot use the latest model. Every valid innocent prompt triggers the policy flag and forces fallback to older, weaker models.

Please fix this incorrect content moderation and stop automatic model downgrade.

What steps can reproduce the bug?

Feedback ID: 01a0bc9e-2e06-7620-9bde-a19dd33e087c

What is the expected behavior?

Normal innocent technical questions should not be falsely flagged for policy violation.
The model should stay on the selected latest model (GPT-6 Astra) instead of automatically downgrading to older, less capable models.

Additional information

This is a false positive content moderation issue. The prompt only contains basic electronics questions with no violating content.
Also there is a config warning: Codex is ignoring 1 unrecognized configuration setting. session-flags: features.thread_tools is ignored.