Tested on Gemini 3.8, Gemini 3.7, Gemini 3.6, Gemini 3.1
OCR and Google API ad GUI Conversion Error Report
Dear Google Employees, I am submitting a bug report. I performed a conversion through the graphical user interface as well as via the API. I was performing an OCR conversion, meaning I converted a PDF file from an image into a text format to generate text output.
Incorrect Tagging of Prohibited Content and Sensitive Words
The system claims that this is prohibited content, which is completely misidentified. There are no profanities, no rude words, no details about sex, and no mention of sexual organs. Yet, the system tells me that it is prohibited content and that sensitive words are used.
I can guarantee you that there are no names of sexual organs in it, so please stop messing around and fix this. Fix it, because it is an absolutely scandalous error.
Processing Failure in Czech and Request for a Fix
The language involved is Czech. I set the OCR conversion to Czech and also requested formatting preservation, yet it failed anyway. So please be so kind, dear Google employees, and fix this, because it is an absolute disaster and it cannot function like this.
ERROR: Processing of file document_cast_%02d-141-150.pdf failed!
Reason for failure: The model returned COMPLETELY EMPTY TEXT!
Exact reason: finishReason: PROHIBITED_CONTENT | finishMessage: The model output could not be generated. This output contains sensitive words that violate Google’s Generative AI Prohibited Use policy. If you think this was an error, send feedback.
Complete candidate response: {“content”: {}, “finishReason”: “PROHIBITED_CONTENT”, “index”: 0, “finishMessage”: “The model output could not be generated. This output contains sensitive words that violate Google’s Generative AI Prohibited Use policy. If you think this was an error, send feedback.”}