Forum post – Gemini API and personal data usage

Gemini API
1

Hello,

We tested the Gemini multimodal API on publicly available document samples (documents found online for testing purposes), and the extraction results were very good.

We would like to understand how Gemini is used in practice when documents contain personal data (PII):

  • Do customers use Gemini to process documents containing personal data?

  • Is it required or recommended to use Vertex AI for such use cases?

  • Are the inputs (images/text) sent to Gemini stored or deleted, and under what conditions?

Our goal is to understand the recommended and intended usage when moving from public test data to real-world documents.

Thank you for your clarification.

3

Hi @ASSU2000_assu
Welcome to the Google AI Forum!!!

Thank you for reaching out!

  • The Gemini API Terms and Usage Policies require responsible use and compliance with all policies, including safety and prohibited content. You should not include sensitive or confidential information when opting in for contributions. Refer this document Data Logging and Sharing

  • Yes, Vertex AI is strongly recommended and in many cases expected. If you want detailed information please refer this Vertex AI and zero data retention Documentation.

  • Further information is available here.