Issue:
OpenAI’s automated abuse and distillation detection appears to be producing false positives for normal developer workflows involving its own products. In at least one case, a flag of this kind was later confirmed by OpenAI as issued in error.
Workflow that may trigger it:
- A developer uses Codex in VS Code to update an app.
- The developer copies Codex’s changes into ChatGPT to review them.
- The developer pastes the same code into a new ChatGPT thread to ask a separate question.