My department handles highly sensitive data, so we are very careful about data leakage.
I have to clarify whether conversation data in Amazon Quick is shared with model provider.
I have confirmed that chat memories are not shared with other users and not used for training or improving underlying LLMs from documents below.
But I am not confident that chat memories are not shared with model provider.
Ask questions, explore data, and get insights with chat in Amazon Quick
Memory and response personalization
This information is securely stored and fetched to enhance your future interactions with chat agents. Memories are specific to you and are not shared with other users.
Data protection in Amazon Quick
Amazon Quick does not use customer data for training or improving underlying LLMs.
Hi @Ryusei_Osera and welcome to the community!
You’ve already confirmed the two key points from Quick’s own documentation, memories aren’t shared with other users, and customer data isn’t used for training or improving LLMs. Regarding whether inference data is shared with the model provider during processing, I wasn’t able to find explicit documentation from Amazon Quick that states this directly. However, Quick’s chat agents are built on Amazon Bedrock, and Bedrock’s documentation explicitly states: “Amazon Bedrock never shares your data with model providers or uses it to train foundation models.”
(Bedrock Security & Privacy). Given that Quick runs on Bedrock, it’s reasonable to conclude that the same data isolation applies, but again, I could not find explicit documentation for this.
Hope this helps.
Thanks, @WLS-Cesar
I figured out Bedrock’s data retention policy.
I have never heard that “Quick’s chat agents are built on Amazon Bedrock”.
Is there any reference page that show that?
