I am trying to create a data store in vertexAI.
The data store will store documents containing confidential information within the company.
However, we are concerned that vertexAI will learn the data stored in the data store and lead to information leakage outside the company.
When creating the data store, there was a checkbox option to “Exclude from generated AI functions”. (I have attached a screenshot in the attached photo. Sorry, it's in Japanese.) I asked Gemini what this does and the response was that if you don't want vertexAI to learn, it is recommended to set it to True. This is related to my concern, and since I am not comfortable trusting only Gemini, I checked the official Google documentation, but there is no clear statement explaining the functionality of this option.
What exactly is the functionality of this option?
Hi @yu_shimada,
Welcome to Google Cloud Community!
From the console, this option means, you will exclude your specific data store from being used for generative content, such as summaries or answer generation. While you are correct about the Exclude from generated AI functions, there is still no public documentation available regarding this.
As discussed in Google Cloud AI/ML Privacy Commitment, Google provides a high level of security and control over how customers use their data for training AI models. It was also mentioned in Google’s Service Specific Terms Section 17 (Training Section), Google won't use your data to train or fine-tune any AI/ML models without your prior permission or instruction. Alternatively, you can review and achieve zero data retention by following these guidelines.
To address the lack of documentation, consider filing an issue report to bring it to the team’s attention. For further clarification, you can contact Google Cloud Support for a one-on-one discussion about Exclude from generated AI functions option.
Was this helpful? If so, please accept this answer as “Solution”. If you need additional assistance, reply here within 2 business days and I’ll be happy to help.
Thanks for your response.
“Google will not use your data to train or fine-tune AI/ML models without your prior permission or instruction.”
I understand that,
Gemini requires an opt-out setting if you do not want your AI models to be trained.
Is there such an explicit setting for vertexAI's data store as well?