DataMasque Synthetically Identical Masked Data (Free 1000 rows)

Databricks SQL Warehouse Support

DataMasque can now mask data held in Databricks SQL Warehouse databases.

IBM Informix Support

DataMasque can now mask data held in IBM Informix databases.

PowerPoint Presentation Masking

The unstructured_text mask now masks Microsoft PowerPoint presentations (.pptx files), extending the existing support for Word documents and PDFs.

Flexible Table Joins

The mask_table joins now support inner and left join type selection, user-defined alias names, and extra on conditions. Aliases allow the same table to be joined more than once and referenced cleanly.

Guaranteed Value Change for From File

The from_file mask gains a force_change option that guarantees the replacement value always differs from the original value.

New Checksum Matcher Types

The checksum matcher in the unstructured_text mask now supports the nz_ird and weighted_dual_checksum types.

AWS Resource Tagging

After a successful masking run on AWS, DataMasque automatically tags the masked RDS, Redshift, DynamoDB, or S3 resource as masked. This is enabled by default on AWS installs and requires new IAM permissions.

PRM Tagging with Your Own IaC

New guidance covers tagging AWS resources for PRM using your own CloudFormation or Terraform, for enterprises that deploy DataMasque with their own infrastructure-as-code on the snapshot-only path.

Database-Tuned Batch Sizes

DataMasque now chooses a default batch size based on the database type, using values tuned through benchmarking for each supported database.

Library Rename Propagation

Renaming a ruleset library now offers to update the rulesets that import it, so they remain valid after the rename.

Stricter Ruleset Validation

Ruleset validation now rejects two previously ignored, undocumented options: metadata_logging on the imitate_uuid mask, and the mask-level context_filter on the unstructured_text mask. Rulesets that set either option must remove it on upgrade. The documented per-matcher context_filters option is unaffected.

Reliable AI Detection in Multibyte Documents

The unstructured_text mask now reliably detects and masks personal data in large documents (over 2,500 characters) that contain multibyte characters such as Chinese, Japanese, Korean text, emoji, or accented characters. This affected Word documents, PDFs, PowerPoint files, and the mask_table task.

Per-Transform Hash Sources in XML

Fixed XML masking so each node_transform uses its own hash_sources. Previously only the first node_transform's hash_sources took effect, producing identical masked values across attributes that should have differed.

Consistent Composite Masking

Fixed composite masks (concat and chain) inside the unstructured_text mask so a value reused across components is masked consistently. Previously only the first component stayed consistent with the rest of the document, so the same last name could be masked to two different values when it appeared both on its own and inside a concatenated email address.

Editor Save Button Fix

Fixed an issue where the Save button did not re-enable after uploading a new ruleset or library file in the editor.