OpenAI account banned for editing config file containing benchmark keywords; appeal ignored
MaxPhoenix_ · reddit · 2026-08-30
An OpenAI user reported having their account deactivated after the safety system mistakenly flagged reference material in a file-edit request.
The Trigger:
- The user asked the model to edit a models. config file.
- A pasted curl command (used as reference) contained keywords from a censorship benchmark (e.g., "crushed beans", toxin terms).
- OpenAI's system flagged the embedded keywords as a "biological" policy violation, despite the user only requesting a file edit.
The Issue:
- Context Failure: The system treated quoted text for an editing task as a direct request for prohibited content.
- Appeal Process: The user's second appeal, which clarified the context, was immediately closed as a "duplicate" without review.
User's Argument:
- Analogized to asking a model to edit a manuscript containing the word "murder" and getting banned as if asking how to commit murder.
- Emphasized the lack of a meaningful correction process for false positives affecting persistent work history.
More from Companies & People
- OpenAI DevDay hands out Codex 'pet' collector cards, attendees show them off — gabrielchua · 2026-10-16
- ML researcher Emtiyaz Khan moves to TU Darmstadt, hiring PhD students via ELISS — EmtiyazKhan · 2026-10-16
- Tripo hosts first world-building hackathon in SF with World Labs and Runway talks — mhdfaran · 2026-10-16
- Dev warns enterprises: OpenAI and Anthropic products are an IP security risk — examachine · 2026-10-16
- a16z podcast: Can AI run the systems after writing the code? — a16z Podcast · 2026-10-16
- Zuckerberg backs electrician training school as US datacenter construction outpaces offices — emmanuelvivier · 2026-10-16