PII Identification (Personal Data Detection) 

PII identification is the process of automatically detecting and classifying personal and sensitive data across systems, databases, and content types. This includes direct identifiers (such as names, email addresses, phone numbers, national IDs) as well as indirect identifiers that can identify individuals when combined. 

Accurate PII identification is a prerequisite for compliant anonymization, pseudonymization, multimodal anonymization, and AI data processing under regulations such as GDPR, HIPAA, CCPA / CPRA, PIPEDA, and India’s DPDP Act. 

How Maya implements PII identification 
Maya uses a combination of AI-based discovery and rule-based classification to identify PII across: 

• SAP and non-SAP enterprise applications (including Oracle, Salesforce, ServiceNow, Microsoft Dynamics, Workday, and Snowflake) 

• Structured data (databases, tables, fields) 

• Semi-structured data (JSON, XML, logs) 

• Unstructured data (PDFs, Word documents, emails) 

• Images, audio, and video files 

The detection works across standard data models and custom structures (for example SAP Z-tables) and can be extended with customer-specific data classes. Identified PII is consistently classified and made available for downstream anonymization, pseudonymization, or AI-safe processing — without manual data mapping. 

 

Anonymization 

Anonymization is the irreversible removal of personal identifiers so that individuals can no longer be identified by any means reasonably likely to be used. 

Proper anonymization can significantly reduce or eliminate regulatory obligations when data can no longer be linked to an identifiable individual, in accordance with applicable laws and regulatory guidance. Depending on jurisdiction and implementation, this may include frameworks such as: 

• GDPR (EU – Recital 26) 

• HIPAA (US – de-identified data under Safe Harbor or Expert Determination) 

• CCPA / CPRA (California – de-identified data) 

• PIPEDA (Canada – risk-based de-identification) 

• India’s DPDP Act (subject to evolving interpretation) 

Anonymized data is commonly used for: 

• Test and QA environments 

• Analytics and reporting 

• AI and machine learning training 

How Maya implements anonymization 
Maya uses Privacy-Enhancing Technologies (PETs) to irreversibly anonymize data while preserving structure, statistical distribution, and cross-system consistency, so that SAP and non-SAP enterprise applications continue to function smoothly on anonymized data. Business processes, validations, integrations, analytics, and AI pipelines remain fully operational. 

 

Multimodal Anonymization 

Multimodal anonymization refers to the ability to anonymize personal and sensitive data consistently across multiple data modalities, not only traditional structured databases. 

This includes: 

• Structured data (databases, tables, fields) 

• Semi-structured data (JSON, XML, logs) 

• Unstructured data (PDFs, Word documents, emails) 

• Images (scans, IDs, photos) 

• Video and audio content 

Multimodal anonymization is critical for modern enterprises because personal data rarely exists in a single format. The same individual often appears simultaneously in ERP systems, CRM platforms, data warehouses, documents, images, recordings, and AI training datasets. 

How Maya implements multimodal anonymization 
Maya applies the same PETs and consistency logic across all data types. Identical identifiers are anonymized deterministically and consistently, regardless of whether they appear in SAP tables, Oracle databases, Salesforce records, ServiceNow tickets, Snowflake datasets, documents, images, or video frames. 

This ensures that: 

• Enterprise applications continue to run reliably on anonymized structured data 

• Documents, files, images, audio, and videos are anonymized without manual intervention 

• Cross-system and cross-format consistency is preserved end-to-end 

• AI and analytics pipelines receive coherent, privacy-safe datasets 

Multimodal anonymization is natively supported across AppSafe™, FileSafe™, and AISafe™. 

​

Pseudonymization 

Pseudonymization replaces direct identifiers (such as names, customer IDs, or national identifiers) with artificial values or tokens. 

Unlike anonymization: 

• Pseudonymization is reversible under strict controls 

• The data remains personal data under GDPR and similar laws 

Pseudonymization is commonly used when organizations need: 

• Controlled re-identification 

• Auditability 

• Consistent identifiers across systems 

How Maya implements pseudonymization 
Maya enables deterministic, cross-system pseudonymization across SAP and non-SAP enterprise platforms (including Oracle, Salesforce, ServiceNow, Microsoft Dynamics, Workday, and Snowflake) without storing raw identifiers in a central vault, ensuring privacy by design and audit readiness. 

 

Data Masking 

Data masking hides sensitive data by replacing it with fictional but realistic-looking values. It is often used to prevent accidental exposure in non-productive environments. 

Important limitations: 

• Masking may preserve patterns or formats 

• Masked data can sometimes be reversed 

• Masking alone often does not meet GDPR, HIPAA, or AI Act requirements 

How Maya uses data masking 
Maya applies masking selectively for display-safe or low-risk scenarios, while using anonymization or pseudonymization where regulatory compliance is required. 

 

De-Identification 

De-identification is an umbrella term describing techniques that reduce the ability to identify individuals in a dataset. 

It includes: 

• Anonymization (irreversible) 

• Pseudonymization (reversible) 

The term is widely used in healthcare, AI governance, and regulatory guidance. 

How Maya supports de-identification 
Maya automatically selects the appropriate de-identification method based on regulation, data sensitivity, business purpose, and target platform. 

 

Re-Identification 

Re-identification is the controlled restoration of original identities from pseudonymized data. 

Typical use cases include: 

• Regulatory audits 

• Legal investigations 

• AI workflows where model outputs must be linked back to real entities 

How Maya enables secure re-identification 
With AISafe™, re-identification happens entirely within the customer’s controlled environment. Combined with Maya’s containerized deployment model, this ensures that no personal data is exposed to external AI models, SaaS providers, or third-party services.