While developing an IM bot using GPT-5.5 Codex, I encountered a false positive GPT cyber policy violation.
What version of Codex CLI is running?
v0.125.0
What subscription do you have?
chatGPT pro 5x
Which model were you using?
gpt-5.5
What platform is your computer?
linux ubuntu LTS 24.02
What terminal emulator and version are you using (if applicable)?
vscode
What issue are you seeing?
I was simply using the GPT-5.5 model through Codex to develop an IM chat bot assistant, and now that development is nearing completion, I asked Codex to test whether the bot functions as intended in my current environment—specifically regarding its ability to respond and assist users. Instead, I encountered this:
ⓘ This chat was flagged for possible cybersecurity risk
If this seems wrong, try rephrasing your request. To get authorized for security work, join the Trusted Access for Cyber
program.
https://chatgpt.com/cyber
What steps can reproduce the bug?
Uploaded thread: 019ddfb5-ef1b-7aa1-aa51-3269b26a8419
Since I uploaded multiple private credentials to GPT for IM bot testing assistance, I had GPT-5.5 mask the records within legal and compliant boundaries before submission, retaining only today's content for the OpenAI team's review.
Please remove the flag from my account.
What is the expected behavior?
Codex GPT-5.5 is working correctly. Please help me test the chatbot to ensure it can accurately read text, view media, and respond based on the chat context.
Additional information
_No response_
3 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Thanks for reporting. This will help us tune our classifier.
Could you please remove the flag from my account? I've read on Reddit that having a flagged account can have serious consequences, and I really don't want to waste time or energy dealing with this through no fault of my own. To be honest, this false positive hasn't just disrupted my development work; it has made me wary of using Codex altogether. I'm worried that any slight misstep might get me flagged again, leading to my requests being rerouted to slower, less capable models, or even getting my account banned.