#7202 500: Error code: 500 - {'error': {'message': 'sensitive words detected (request id: 20260912103530480575230jgbvwa9khcDJ5

A 500 error with sensitive_words_detected is returned by the AI provider/API gateway itself (a new_api_error type), not by pyVideoTrans. It means the text sent for translation was flagged and rejected by that service's content-filtering system, so the request never completed and OpenAI's client raised InternalServerError.

What to do:

1. Identify the culprit text. The filter is triggered by specific subtitle content (profanity, political terms, violence, sexual words, etc.). Manually inspect the source SRT of the batch that failed.

2. Switch or use a different AI translation channel. Some relay/gateway providers (the new_api_error type often indicates a third-party/proxy OpenAI-compatible endpoint) enforce stricter moderation than the official API. Try:

   • a different AI channel (e.g. DeepSeek, Gemini, or the official OpenAI endpoint),
   • or a non-AI channel such as Google(免费) / M2M100(内置).
3. Reduce the chance of re-triggering. Go to Menu → Tools/Options → Advanced Options → Subtitle Translation and set AI translation batch size (lines per request) to 1, so only one line is sent at a time; a single offending line can then be isolated and manually edited.
4. Pre-edit the source SRT. Rewrite/replace the flagged words in the source subtitle, then re-run translation (do not tick 清理已生成 if you want to keep the existing transcription).
5. If it is caused by the prompt, not the content, note that when "Send full SRT to AI" is enabled the whole subtitle block is transmitted in one request — disabling it and translating line-by-line usually avoids mass rejection.

Also note your version is v3.97, which is quite old; upgrading is recommended before further troubleshooting.

Please refer to the related documentation:

• FAQ / Common problems
• Translation channels
• Advanced options
• Why do blank subtitle lines appear after translation?