Regulatory Mapping

APP is designed with regulation in mind. This page maps APP features to specific regulatory requirements, with a focus on the EU AI Act.

EU AI Act Article 50

The EU AI Act’s transparency obligations become enforceable on 2 August 2026.

Compliance matrix

Obligation	Article	APP Feature	Coverage
Mark output as AI-generated	50(2)	`ai_generated: true` in `_ai_provenance` metadata	Full
Machine-readable, detectable format	50(2)	Three embedding modes	Full
Free detection mechanism	50(2)	Public verification endpoints	Full
Publicly available interfaces	50(2)	HTTPS verification, no auth	Full
Deployer disclosure	50(4)	`ai_generated` field for downstream use	Full
Human review exemption	50(4)	`review.human_reviewed` + timestamp	Full
Non-breaking marking	50(2)	`_ai_provenance` key is ignorable; headers are non-breaking	Full

How to achieve compliance

Embed APP metadata in all AI-generated output
Expose a verification endpoint (recommended)
Record human review when it occurs
Use non-breaking embedding (inline JSON or HTTP headers)

Other jurisdictions

Jurisdiction	Framework	APP Relevance
US	Executive Order 14110	Provenance tracking aligns with EO guidance
US	NIST AI RMF	Supports Transparency and Accountability functions
China	Interim Measures for Generative AI	APP’s `ai_generated` field satisfies labelling requirements
Canada	AIDA (proposed)	Ready-made transparency mechanism
UK	AI Safety Institute guidance	Concrete implementation of voluntary best practices
Global	ISO/IEC 42001	Provides required audit trail and documentation

Compliance checklist

Embed APP metadata in all AI-generated output
Set ai_generated: true for all AI-generated content
Include accurate generator information
Record generated_at timestamps
Generate unique generation_id for each generation
Expose a public verification endpoint
Implement human review tracking when applicable
Include content_hash for integrity verification
Validate metadata against the JSON Schema

Last updated on March 6, 2026

MCP Integration SDKs