Does anyone have tips to get rid of the endless possible content warning violations? I tried to use ChatGPT to improve my intake notes as a psychiatric provider. However, the notes always get flagged as possible content violations. I always give a thumbs down and explain why I am not violating any policies (no personal patient identifiers are used, no HIPAA violations). I assume it is because of the nature of most of the topics discussed (many are trauma-based). However, it continues to be an ongoing problem with no improvement. I get the possible content warning with everything and ChatGPT will not generate any content. What is the point in giving feedback if nothing changes? It’s been a few months. I am a paid member.
Any advice appreciated! Thanks!
Hi @foreverlaur
Welcome 
to the community!
The issue you’re encountering likely stems from the sensitive nature of the psychiatric topics in your intake notes, which trigger OpenAI’s content moderation systems. These systems are designed to avoid generating harmful or disallowed content, particularly around subjects like trauma, mental health, or self-harm, which are sensitive and could be flagged as unsafe by the model’s classifiers.
OpenAI’s models, including ChatGPT, have strict safety and content moderation mechanisms in place to prevent the generation of potentially harmful content. Trauma-based topics often involve discussions around violence, abuse, or other distressing content. Even if your prompt is intended for professional purposes and doesn’t violate any guidelines, the system may still flag it due to its automatic sensitivity to content that could be interpreted as distressing or harmful.
We have also some experience like you. My kid’s class teacher has given a presentation about mineral and rocks. It is a important topic in curriculum. Words like “mineral” can sometimes be flagged when the system associates them with hazardous or industrially sensitive materials for example substances used in weapons or toxic materials. If the model doesn’t fully understand the context of a benign educational query, it might mistakenly categorize it as something riskier.
But we solved like that; we rephrased some questions, and also when we ask we used some placeholders or triple quotation marks, for example;
“My kid’s class teacher has given homework about [topic here]…” in brackets like placeholders, or in triple quotation marks “”“topic here”“”.
It was solved, and we did not get flag. I am not sure if it helps about your topic.