EnforceCore EnforceCore

GDPR — Compliance Mapping

EnforceCore v1.0.15 · Last updated Feb 2026
Info

Info: This document maps EnforceCore capabilities to relevant GDPR articles. It is not legal advice and does not constitute a compliance certification. Organizations must perform their own data protection impact assessments (DPIAs) with qualified legal counsel.

Scope

The General Data Protection Regulation (GDPR) applies to any organization processing personal data of EU/EEA residents. When AI agents process personal data — names, emails, phone numbers, health records — GDPR imposes strict requirements on data minimization, transparency, and accountability.

EnforceCore helps address these requirements at the runtime enforcement layer — the point where agent actions are intercepted before execution.

Article-by-Article Mapping

Article 5 — Principles Relating to Processing

Principle GDPR Requirement EnforceCore Capability
Lawfulness, fairness, transparency Processing must be lawful and transparent Merkle-chained audit trail provides full transparency of all agent actions
Purpose limitation Data collected for specified purposes only Policy engine restricts tool access to explicitly allowed operations
Data minimization Only necessary data should be processed PII redaction removes unnecessary personal data from agent outputs
Accuracy Personal data must be accurate Deterministic policy evaluation — no model-dependent decisions on data handling
Storage limitation Data kept only as long as necessary Audit logs are append-only; retention policies can be configured
Integrity and confidentiality Appropriate security measures Fail-closed enforcement, tool gating, network controls

Article 25 — Data Protection by Design and by Default

Requirement EnforceCore Capability
Implement appropriate technical measures PII redaction is enabled by default when configured in YAML policy
Data protection integrated into processing EnforceCore sits at the runtime boundary — personal data is intercepted before it leaves the agent
Only necessary personal data processed pii_redaction.categories specifies exactly which PII types to detect and redact
Default settings protect personal data on_violation: block is the recommended default — violations are blocked, not logged

Article 30 — Records of Processing Activities

Requirement EnforceCore Capability
Maintain records of processing activities Merkle-chained audit trail records every agent action with timestamps
Include purposes, categories of data, recipients Audit log entries include: tool name, arguments, policy evaluation result, redaction actions
Records available to supervisory authority on request enforcecore verify CLI command validates audit trail integrity; logs can be exported

Article 32 — Security of Processing

Requirement EnforceCore Capability
Appropriate technical and organizational measures Runtime enforcement layer with fail-closed design
Pseudonymization and encryption PII is replaced with placeholders ([REDACTED-EMAIL], [REDACTED-SSN]) before leaving the agent boundary
Ensure ongoing confidentiality, integrity, availability SHA-256 Merkle chain ensures audit trail integrity; policy engine ensures consistent enforcement
Regular testing and evaluation enforcecore dry-run allows testing policies without executing actions; 1,500+ automated tests

Article 35 — Data Protection Impact Assessment (DPIA)

Requirement EnforceCore Capability
Assess impact of processing on data protection EnforceCore's audit trail provides the data needed for DPIAs
Describe the processing and its purposes Policy YAML files document exactly what the agent is allowed to do
Assess necessity and proportionality Tool allow/deny lists enforce the principle of least privilege
Measures to address risks PII redaction, rate limiting, cost guards, network controls

PII Categories Supported

EnforceCore's regex-based PII detection currently supports:

Category Example Pattern Redaction Output
Email user@example.com [REDACTED-EMAIL]
Phone +1-555-0123 [REDACTED-PHONE]
SSN 123-45-6789 [REDACTED-SSN]
Credit Card 4111-1111-1111-1111 [REDACTED-CC]
Name John Smith [REDACTED-NAME]
Address 123 Main St, City, ST 12345 [REDACTED-ADDRESS]
Warning

Warning: Regex-based PII detection has inherent limitations. It may not catch all PII variants (e.g., obfuscated data, uncommon formats). Organizations handling sensitive data should layer additional detection methods (NER models, custom patterns) alongside EnforceCore.

Gaps and Limitations

Area Status Notes
Right to erasure (Art. 17) ⚠️ Partial Audit logs are append-only by design (integrity). Erasure of personal data from logs requires organizational procedures.
Data portability (Art. 20) ⚠️ Partial Audit logs are exported as JSONL — portable but not in a standardized personal data format.
Consent management ❌ Out of scope EnforceCore does not manage user consent. This must be handled by the application layer.
Cross-border transfer (Art. 46) ❌ Out of scope EnforceCore is infrastructure — deployment geography is the organization's responsibility.
DPO notification ❌ Out of scope Organizational responsibility, not a technical control.

Recommended Configuration

For GDPR-aligned deployments, we recommend:

name: "gdpr-compliant-policy"
version: "1.0"

rules:
  pii_redaction:
    enabled: true
    categories: [email, phone, ssn, credit_card, name, address]
    strategy: placeholder

  allowed_tools:
    - search_documents
    - generate_summary

  denied_tools:
    - send_email
    - write_file
    - execute_shell

on_violation: block

Combined with:

  • enforcecore verify in CI/CD to validate audit trail integrity
  • Regular DPIA reviews using exported audit logs
  • Organizational consent management upstream of EnforceCore

See Also