Security, privacy, and compliance for Gemini in BigQuery

This document describes the controls that support the security ofGemini in BigQuery. These controls can also helpyou meet the privacy and regulatory requirements that apply to your business.Gemini in BigQuery is built on Google Cloudinfrastructure. Your data remains in your control. For more information, seeService Specific Terms.

The following controls apply to Generally Available (GA) Geminiin BigQuery features:

• Your data is not used for training models without your permission.Google does not use your prompts, responses, or schema information to trainits models unless you explicitly opt in.
• Your BigQuery data remains within your chosen location.Gemini in BigQuery respects yourBigQuery data at rest data residency settings. The coreBigQuery engine that runs queries and stores your datacontinues to honor your location constraints. For more information, seeHow Gemini in BigQuery processes data.
• Gemini in BigQuery is covered by Googlesecurity and compliance offerings. Coverage includes certifications likeSOC 1/2/3, ISO/IEC 27001, andHIPAA compliance. For more information, seeGoogle security and compliance offerings.

The security, privacy, and compliance for Google Cloud services are ashared responsibility.Google secures the infrastructure that Google Cloud services run on, and itprovides you with tools such as access controls to let you manage who has accessto your services and resources. For more information about how the infrastructure is secured, seeGoogle infrastructure security design overview.

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 responsibleAI.

Gemini in BigQuery architecture

The following diagram shows the components of the Gemini inBigQuery architecture.

Important: Gemini in BigQuery processes data in theUS orEU jurisdictions where the data resides. Data outside these jurisdictions is processed globally. To learn more about where Gemini in BigQuery processes your data, seeWhere Gemini in BigQuery processes your data.

How Gemini in BigQuery processes data

When a user uses Gemini in BigQuery, a prompt andits relevant context are sent to Google's large language models (LLMs) forprocessing. Google manages the specific models used to generateGemini in BigQuery responses.

1. Prompt. A user enters a prompt as a natural language question, suchas "Show me the top 5 customers by sales last quarter". Or, a user types apartial SQL or Python snippet in the Google Cloud console inBigQuery Studio with Gemini in BigQueryenabled.
2. Contextualization. Gemini in BigQueryaccesses the relevant metadata and schema of your BigQuerytables to add context to the user's prompt. Contextual information can includesampling data from tables and job histories. Gemini inBigQuery only has access to the resources to which the userhas access.
3. Gemini processing. The prompt and contextualinformation are sent to Gemini's LLMs for processing.Gemini in BigQuerydoesn't retain or store contextual information. Gemini inBigQuery uses the existing BigQuery contextthat is stored in Dataplex Universal Catalog and Spanner.This information resides in the same location as your data.Gemini generates a response, such as a SQL query, a datainsight, or a Python code snippet.
4. Response. The response is returned to the BigQueryinterface. The user can then run the generated code, modify it, or continueto iterate on the response by using Gemini. You can providefeedback from Gemini in BigQuery in theGoogle Cloud console. To learn more about providing feedback, seeProvide feedback.

Security controls

Gemini in BigQuery uses the security controls ofGoogle Cloud to help protect your data and resources. These controlsinclude the following:

• Authentication. Users authenticate by using theirGoogle Cloud credentials, which can be integrated with your existingidentity provider.
• Access controls. You can use Identity and Access Management (IAM) to controlwho has access to Gemini in BigQuery and whatactions they can perform.
• Network security and VPC-SC. Gemini inBigQuery traffic is encrypted in transit and at rest. Youcan also useVPC Service Controls tocreate a security-enhanced perimeter around yourBigQuery resources.

Data protection and privacy

Gemini in BigQuery is designed to protect theprivacy of your data. Google's privacy policies and commitments apply to alldata processed by Gemini in BigQuery.

• Data encryption. Your data is encrypted at rest and in transit.
• Data access. Google personnel have limited and audited access toyour data.
• Data residency. Your BigQuery data-at-restis stored and processed in the Google Cloud region you select.

Certifications and capabilities

Generally available (GA) Gemini in BigQueryfeatures are covered by the certifications and security statements ofGemini for Google Cloud with exception of thefollowing limitations:

• Gemini in BigQuery doesn't provide dataresidency for individual locations. Gemini processing can bespecified for data with theUS- andEU-supported jurisdictions. Dataoutside these jurisdictions is processed globally. To learn more, seeWhere Gemini in BigQuery processes yourdata.
• Cloud logging audit logs are not available for Gemini inBigQuery user prompts and responses.
• Gemini in BigQuery is not included insupportedAssured Workload packages.

To learn more about certifications and security for Gemini forGoogle Cloud, seecertifications and security for Gemini forGoogle Cloud.

Secure and responsible use

You should adhere to the following best practices to help ensure the secure andresponsible use of Gemini in BigQuery:

• Use IAM to give the least privilege necessary. Forinformation about security best practices in BigQuery, seeIntroduction to security and access controls inBigQuery.
• Be mindful of the data you include in your natural language prompts inBigQuery, such as sensitive or personal information.
• Review and validate the responses generated by Gemini inBigQuery. Always treat AI-generated code and analysis assuggestions that require human review.
• Only enable Gemini in BigQuery forprojects that don't require compliance offerings other than those listedpreviously and byGemini for Google Cloud.For information about how to turn off or prevent access toGemini in BigQuery, seeTurn off Gemini in BigQuery.

What's next

• ReadOverview for Gemini in BigQuery.
• ReadSetup forGemini in BigQuery.
• ReadIntroduction to data governance inBigQuery.
• ReadCertifications and security for Gemini forGoogle Cloud.
• ReadSupported products by control package.
• ReadHow Gemini for Google Cloud uses your data.