Privacy Vault

PII Detection & Performance Enhancements

This release focuses on improving PII identification accuracy, Agentic workflow performance, and processing efficiency.

Enhanced PII Identification

Improved PII detection for JSON and tabular content provided as string payloads, particularly in Agentic workflows.

Improved Agentic Throughput

Optimized Agentic prediction processing for higher throughput and better performance under load.

Faster Processing

Reduced processing latency through performance optimizations.

Optimized Request Processing

Introduced asynchronous execution and dynamic batching to improve processing efficiency and GPU utilization.

Improved Structured Data Handling

Reduced incorrect flagging of non-sensitive fields in Flat and Nested JSON structures.

Improved Tabular Data Detection

Enhanced PII identification in tabular text containing row and column structures.

Expanded PII / PHI / PCI Entity Support

Added support for new sensitive data entities across healthcare and regional compliance use cases.

Newly Added Entities

• BIOMETRIC_DATA
• BLOOD_TYPE
• MEDICAL_CODE
• MEDICAL_CONDITION
• MEDICATION
• MEDICAL_PROCEDURE
• PH_TIN
• GSTIN
• AADHAAR
• UK_UTR
• UK_NIN
• US_ITIN
• CANADIAN_SIN

Platform Optimization

Vault Table Index Optimization

Implemented vault table index optimization to improve:

• Storage efficiency
• Infrastructure cost optimization
• Database performance and scalability

Security

Vulnerability Fixes for Patch

• Applied security patches to address known vulnerabilities

Compatibility

• No breaking changes
• Existing APIs and workflows remain fully compatible

Multilingual API Support

The upcoming April release introduces multilingual support across APIs, enabling broader usability across global teams and applications.

What’s new:

• API requests and responses will support multiple languages
• Improved handling of non-English inputs for masking, unmasking, and processing workflows
• Consistent performance across supported languages

Supported Languages:

• English
• German
• French
• Spanish
• Dutch
• Arabic
• Japanese

Impact:

• Enables seamless adoption in multi-region deployments
• Improves accuracy for PII detection and processing in native languages
• Reduces dependency on pre-translation before API calls

Supported entities

See the full list of supported entities in PII / PHI / PCI entities.

2 more entities have been added:

• NATIONAL_ID
• PASSWORD

Note:** CITY, COUNTRY**, and PINCODEare derived sub-entities extracted from the existing ADDRESS entity. They provide more granular location data without requiring separate detection logic.

Vault & Tokenization

• Prefix and suffix support fully generalized across all entities, including custom and auto-detected PII
• Policy-driven masking behavior made consistent across sync mask, async mask, and unmask
• Improved token consistency guarantees across retries and async jobs

Policies & Metadata

• Metadata updates now propagate immediately to all new masking operations
• Clear separation between anonymizable entities, pseudonymizable entities, and excluded entities

Security & Admin

• Stronger validation on namespace deletion and background job tracking
• Improved error messages for expired or invalid admin tokens

Vault & Tokenization

• attributes object added to Mask and Unmask APIs — enables tracking of page number, source ID, or custom metadata
• Attribute values preserved through async workflows

Reporting & Compliance

• Audit logs enhanced to include attributes when present
• Better pagination stability for large audit datasets

Data Scan & Discovery

• Scan Details API now returns identified percentage per entity and ML-identified vs user-defined PII separation
• Pagination support added for large schema scans

Policies & Metadata

• PI conclusion percentage now configurable globally or per object

Vault & Tokenization

• Identify-and-mask enhanced with individual token position tracking and expanded toxicity analysis
• Unmask API now consistently returns toxicity scores

Data Scan

• Async scan status lifecycle stabilized: PENDING, IN-PROGRESS, SUCCESS, FAILED

Security & Admin

• Super Admin APIs hardened with auth token lifecycle enforcement and explicit permission checks (mask, unmask, metadata_change)
• Namespace creation and deletion moved fully to background jobs

Policies & Metadata

• Policy name support added to Mask and Unmask APIs
• Default policy behavior clarified when policy_name is omitted

Vault

• Better handling of mixed-content payloads during auto-detect masking

Vault & Tokenization

• Async Mask and Async Unmask introduced
• Async Status API added with detailed job lifecycle and timestamps

Vault

• Unified Mask API for mask with token, mask with format, and identify-and-mask (auto-detect)
• UTF-8 support across all APIs

Policies & Metadata

• Namespace-level policies introduced
• Support for anonymization vs pseudonymization

Data Scan

• Initial release of Data Scan APIs (subscription only)

Reporting & Compliance

• Namespace-level audit summaries introduced
• Usage summary APIs added for billing visibility

Security

• Improved authentication error handling
• Clear differentiation between expired token, invalid token, and missing authorization header

Data Scan

• Object-level scan support (database, schema, table)
• Scan object listing APIs introduced

Vault

• Format-preserving masking expanded
• Additional financial and healthcare entities supported

Vault

• Toxicity detection introduced
• Toxicity scoring added to masking responses