This chat was flagged for possible cybersecurity risk

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

• #20694
• #20632
• #20803
• #20234
• #19403

Powered by Codex Action

Thanks for reporting. This will help us tune our classifier.

Got the same issue

ⓘ 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

I was working on a agentic chatbot app, particularly on a RAG flow where I asked to debug and trace a flow with below prompt

Lets debug and trace the full rag flow, like questions, see if analyser is properly adding the questions and flow is correct, at last the respond should be getting context that we did not find any information on asked question, so respond agent should naturally say it does not have information on it, not inventing the info.

It is extremely annoying. Literally every second prompt which I am doing is triggering this annoying productivity killer. I am crafting a prompt for 30mins expecting codex to go ahead and start research and coding and leave my workplace for an hour to just come back and see that it stopped processing the prompt after the first 3mins... Definitely one of the worst things brought up my OpenAI and while I was considering to fully switch to codex/openai, I am now rather going back to Anthrophic as this cybersecurity risk alert is triggering on toooo many things!

Honestly, that's like banning kitchen knives for slicing bread just because they could be used to hurt someone.
Codex is literally just a tool...
The authorities should go after the people who misuse it, not the tool makers.

Hey folks, are we planning to resolve this? Even if you have flagged someone, can you at least set a timer of 30 minutes - 2 hours to investigate usage and then reset the flag? This is not just annoying, but this is a waste of time/resources.

I'm getting flagged every single prompt for doing basic tasks on my own VPS editing spreadsheets. Absolutely infuriating

Your conversations have multiple flags for possible cybersecurity risk. Responses may take longer because extra safety checks are on. To get authorized for security work, join the Trusted Access for Cyber program: https://chatgpt.com/cyber

for changing a simple button color.

aren't they resolving it ..... why they think that we are hacking some kind of national treasure website .... somebody please resolve this .... cant even code a simple blob code

I was reporting this via the support bot on the site. It seems like I was banned by the damn tin can after I told it "either make it work or refund my pro subscription fee". It has just stopped responding. I'm just doing a backup of my old hard drives if it matters...

Also Pro+ Sub same issue for trying to ssh into a Raspberry Pi and also configuring and ESP32 IOT device to water my plants. For the prices I'm paying, this should not be an issue. This is wild. Please fix this, like another user said it's a kitchen knife. Stop it!

I liked Codex' help for my own IoT devices inside MY own home, using MY own local network with MY own IoT devices, connected to the ISP that I am paying monthly.

Codex:
’ll do this as research only: S340-specific streaming support, official Eufy docs, and community/API bridge options. I’ll separate confirmed facts from likely/inferred paths so we don’t build around wishful thinking.

Searching the web 2 times
Searching the web 2 times
This content 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 

Apparently asking it to find/research OFFICIAL documentation is being seen as "cybersecurity risk" ?
If this joke doesn't get fixed sooner or later, I am (unfortunately) forced to cross over to other providers.

I liked Codex' help for my own IoT devices inside MY own home, using MY own local network with MY own IoT devices, connected to the ISP that I am paying monthly. `` Codex: ’ll do this as research only: S340-specific streaming support, official Eufy docs, and community/API bridge options. I’ll separate confirmed facts from likely/inferred paths so we don’t build around wishful thinking. Searching the web 2 times Searching the web 2 times This content 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 `` Apparently asking it to find/research OFFICIAL documentation is being seen as "cybersecurity risk" ? If this joke doesn't get fixed sooner or later, I am (unfortunately) forced to cross over to other providers.

<img width="328" height="480" alt="Image" src="https://github.com/user-attachments/assets/58dd95eb-5fd0-4d91-a0b7-3ba92be5abd3" />

They Keep closing everyone that is posting the same issue, as closed duplicate. It's more like were just be dismissed and rather then show the real staggering number of users with the same BS flags we just dismissed, BUT it's been over a month of your premium service, i expect a premium e4xperince or please refund this negative experience. Honestly, you're going to flag for this !@#$ when Anthropic is winning the cyber security researcher crowd? Aren't they the super safe goal orientated ones? I don't see Claude getting his panties in a bunch when I want to setup pi-hole dns. @etraut-openai can you tune the damn thing down already?!