This document describes how Gemini for Google Cloud, which offers AI-powered assistance,conforms toGoogle's privacy commitmentwith generative AI technologies. When you use Gemini in theGoogle Cloud console or in a development environment, Google Cloudhandlesyour prompts in accordance with ourterms ofservice andCloud Data Processing Addendum.

For more information about Gemini for Google Cloud, seeGemini for Google Cloud overview.

Google's privacy commitment

Google was one of the first in the industry to publish anAI/ML privacycommitment,which outlines our belief that customers should have the highest level ofsecurity and control over their data that's stored in the cloud. That commitmentextends to Google Cloud generative AI products. Google helps ensure thatits teams are following these commitments through robust data governancepractices, which include reviews of the data that Google Cloud uses in thedevelopment of its products. You can find more details about how Googleprocesses data inCustomer Data Processing Addendum (CDPA)or the dataprocessing agreement applicable to your Google Cloud service.

Data you submit and receive

The questions that you ask Gemini, including any inputinformation or code that you submit to Gemini to analyze orcomplete, are calledprompts. The answers or code completions that you receivefrom Gemini are calledresponses.

Gemini doesn't use your prompts or its responses as data totrain its models. Some features are only available through theGemini for Google Cloud Trusted Tester Program,which lets you optionally share data, but the data is used for productimprovements, not for training Gemini models.

Code customization letsyou get code suggestions based on your organization's private codebase directlyfrom Gemini Code Assist. When you use code customization, wesecurely access and store your private code. This access and storage isessential for delivering the code customization service you've requested. Toconfigure and use code customization, seeConfigure and use Gemini Code Assist code customization.

Because Gemini is an evolving technology, it can generate outputthat's plausible-sounding but factually incorrect. We recommend that youvalidate all output from Gemini before you use it. For moreinformation, seeGemini for Google Cloud and responsible AI.

Encryption of prompts

When you submit prompts to Gemini, your data is encryptedin-transit as input to the underlying model in Gemini. For moreinformation on Gemini data encryption, seeDefault encryption at restandEncryption in transit.

Program data generated from Gemini

Gemini is trained on first-party Google Cloudcode as well as selected third-party code. You're responsible for thesecurity, testing, and effectiveness of your code, including any codecompletion, generation, or analysis that Gemini offers you.

Gemini also provides source citations when suggestions directlyquote at length from a source to help you comply with any license requirements.

Because responses in Gemini are generated from a model that'strained on many lines of code, you should exercise the same care withGemini-provided code that you would with any other code. Makesure that you test the code properly and check for security vulnerabilities,incompatibilities, and other potential issues.

What's next

• Learn about thesecurity, privacy, and compliance ofGemini Code Assist.