LLMMedium
Users Bypassed ChatGPT's Content Filters with Ease
- Occurred
- Nov 30, 2022
- Published
- Mar 6, 2023
- Source
- AI Incident Database
Summary
Users reported bypassing ChatGPT's content and keyword filters with relative ease using various methods such as prompt injection or creating personas to produce biased associations or generate harmful content.
Developers
- OpenAI
Deployers
- OpenAI
Harmed parties
- ChatGPT users
- OpenAI
Classification
Category confidence: 95%
Severity confidence: 90%
Rationale: ChatGPT content filter bypass via prompt injection is a language model security failure.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8