You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Codex] Repeated cyber_policy blocks during authorised migration regression testing #56615
What version of the Codex App are you using (From “About Codex” dialog)?
v0.164.0
What subscription do you have?
Plus
What platform is your computer?
Linux 7.0.0-38-generic x86_64 unknown
What issue are you seeing?
Codex is repeatedly interrupting regression tests for a large migration in my own repository (PR #416).
The task involves crash recovery, retry behaviour, authentication, synchronisation and compatibility checks. The agent had already confirmed a crash-and-retry failure and was investigating additional regressions.
Observed behaviour
Model: GPT-6.1 Sol, High reasoning
Date: 24 October 2026
Failures: 00:57 and 01:02 BST
Error: "This content was flagged for possible cybersecurity risk."
Three subagents were running when the failures occurred.
The failure stops legitimate verification and provides no useful information about which operation caused the classification.
I also attempted to contact OpenAI Support, but the automated support assistant did not provide a usable escalation to a human.
Requested action
Please investigate the affected safety classifications, identify what triggered them and provide a supported way to continue authorised regression testing.
Please also explain how customers can obtain human review when automated safety systems repeatedly block legitimate development.
Activity
github-actions commentedon Oct 24, 2026
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action