AI Governance - February 20, 2026 - 202602.1.0 (Winter Release)

Embed governance directly into AI development workflows to automatically align controls across related frameworks, track compliance continuously, and apply consistent, measurable guardrails across AI systems built in Databricks and beyond. This powerful enhancement allows customers using Databricks to pursue compliance and best practices the Databricks ecosystem for secure and responsible AI development.

Enable systemic risk visibility across linked inventory objects in AI Governance. When enabled, parent objects (such as AI Systems or Projects) inherit risks from related child entities (such as Models or Datasets), allowing organizations to automate comprehensive risk views.

For more information, see Risk Aggregation for Related AI Records. This feature is related to the following idea: PDGI-I-1706

Enable respondents to add multiple relationships in an AI Governance questionnaire. This enhancement allows respondents to provide all required context and information about AI activities in their organization when completing an assessment.

For more information, see Launching Assessments for AI Governance

|Available with Solution Packaging Only

Establish standardized global records for AI systems, models, and datasets while maintaining contextual, local records across business units, use cases, and lifecycle stages. This capability lets teams track AI asset evolution through a clear audit trail, enforce standardized attributes at the global level, and manage jurisdiction-specific or use case-specific requirements at the local level.

For more information, see AI Governance Global/Local Feature Guide. This feature is related to the following ideas: AI-I-31, PDGI-I-1416

|Available with Solution Packaging Only

Automatically scan cloud environments, APIs, pipelines, and applications to detect active AI models and agents in your Amazon Bedrock accounts. This feature captures metadata including model type, owners, input/output behavior, and associated datasets, providing a real-time, searchable inventory that serves as the foundation for governance policies, risk assessments, and audit trails.

For more information, see AI Automation Feature Guide.

|Available with Solution Packaging Only

Automatically scan cloud environments, APIs, pipelines, and applications to detect active AI models, agents, and datasets in your Databricks accounts. This feature captures metadata including model type, owners, input/output behavior, and associated datasets, providing a real-time, searchable inventory that serves as the foundation for governance policies, risk assessments, and audit trails.

For more information, see AI Automation Feature Guide.

|Available with Solution Packaging Only

Automatically scan cloud environments, APIs, pipelines, and applications to detect active AI models and agents in your Google Cloud accounts. This feature captures metadata including model type, owners, input/output behavior, and associated datasets, providing a real-time, searchable inventory that serves as the foundation for governance policies, risk assessments, and audit trails.

For more information, see AI Automation Feature Guide.

|Available with Solution Packaging Only

Automatically scan cloud environments, APIs, pipelines, and applications to detect active AI models in your SageMaker accounts. This feature captures metadata including model type, owners, input/output behavior, and associated datasets, providing a real-time, searchable inventory that serves as the foundation for governance policies, risk assessments, and audit trails.

For more information, see AI Automation Feature Guide.

|Available with Solution Packaging Only

Automatically scan cloud environments, APIs, pipelines, and applications to detect active AI models and agents in your Azure Foundry accounts. This feature captures metadata including model type, owners, input/output behavior, and associated datasets, providing a real-time, searchable inventory that serves as the foundation for governance policies, risk assessments, and audit trails.

For more information, see AI Automation Feature Guide.

Continuously scan, detect, and catalog AI models and agents across your AWS-native ecosystem. This seamless integration experience enables real-time visibility into your organization's AI landscape, mapping relationships between models, training data, and business processes to strengthen risk management and compliance. By eliminating manual tracking, teams can focus on innovation while maintaining comprehensive oversight of their AI systems through automated discovery and continuous monitoring.

For more information, see Automated AI Discovery: Amazon SageMaker Integration for AI Governance Models

Discover and catalog AI models, notebooks, and experiments across Google Cloud while maintaining centralized governance controls. This seamless integration experience enables real-time visibility into your organization's AI landscape, mapping relationships between models, training data, and business processes to strengthen risk management and compliance. By eliminating manual tracking, teams can focus on innovation while maintaining comprehensive oversight of their AI systems through automated discovery and continuous monitoring.

For more information, see Automated AI Discovery: Google Vertex Integration for AI Governance Models.

|Available with Solution Packaging Only

Improve data use governance functions by automating dataset evaluation and surfacing metadata that clarifies data quality, sensitivity, and readiness. This accelerates access request reviews, ensures responsible data usage, and helps teams build AI models on compliant, trustworthy datasets, which enhances both accuracy and oversight.

Create a report of all records of a specific inventory and the columns eligible for bulk import, or the current saved view of an inventory screen including applied filters. This provides you with two flexible reporting methods to gather comprehensive insights into inventory records.

For more information, see Exporting AI Inventory Records.

Use of the same attribute in multiple actions is now supported in AI Governance workflow stage rules. Previously once an attribute was used in an action within a stage rule, it was no longer available for other stage rules.

System attributes now indicate if theResponse typeis Single select or Multi select so users clearly understand if one or more options can be selected. This is available in the list screen view and details view of the attribute.

Introduced a new column calledList typefor users to more easily find Single select and Multi select options for User and Location lists. This column is also available as a filter option.

Added related entity support in AI Governance column reports to report the following objects:

• Assessment Summary

• Assets

• Vendors

• Entities

• Projects

• Models

• Datasets

• AI Systems

Automatically fill out assessments based on past assessments or scanned documents with the AI auto-responses for assessments feature set, which includes AI Doc Scanner and AI Inventory Analysis.

This change was introduced in the Winter release and communicated for Assessment Automation in the release notes, but it also applies to AI Governance and has been retroactively added to the release notes. Restrictions apply for some AI Governance inventory.

For more information, see Leveraging AI Auto-Responses for Assessments.

Issues that may impact our customers are documented in our Known Defects Database located here with approved workarounds and targeted releases for resolving the issue.