Clean Rooms

Bringing real-time fraud prevention to government benefits

Databricks is helping federal agencies move fraud detection for benefit programs from after-the-fact reviews to real time, building on techniques honed with banks, credit card companies, and insurers. With over 80% of federal executive departments already on the platform, tools like OpenSharing and Clean Rooms let agencies share fraud signals across silos without exposing raw data, surfacing schemes no single agency could catch alone.

The SDK adds support for new clean room task run management and PostgreSQL CDF configuration, plus numerous new fields across compute, ML, pipelines, and serving services. Breaking changes include IAM v2 method endpoint updates, type changes for internal IDs, and removal of deprecated fields in jobs and ML services.

Clean Rooms now supports listing task runs and JAR analysis reviews, alongside new configuration fields for Azure capacity reservation groups and pipeline source metadata columns. In a breaking change, the codeSourcePath field has been removed from AiRuntimeTask.

This release adds SDK support for JAR analysis assets, review workflows, and task run listing in Databricks Clean Rooms, alongside a new JDBC connection type for catalog integrations. Additional updates introduce Azure capacity reservation group settings for compute instance pools, initial parent paths for bundle deployments, and lifetime attributes for ML time windows.

The Clean Rooms API now supports shared output settings, notebook asset metadata, and collaborator aliases. Pipeline configurations for Google Ads, Meta Marketing, and TikTok Ads now include custom report options.

This release adds shared output options, notebook metadata, and collaborator alias fields to the Clean Rooms Java SDK. It also introduces custom report options for Google Ads, Meta Marketing, and TikTok Ads ingestion pipelines.

How Stagwell built privacy-safe ID matching on Databricks

Stagwell built a privacy-safe ID matching solution on Databricks, leveraging Databricks Clean Rooms and Marketplace apps to securely match first-party data with identity graphs. This enables brands to create actionable audiences without exposing sensitive information or raw records.