Transparency programme

AI Labelling

Proposals for standardised, meaningful and machine-readable disclosure of AI involvement.

Why label Artificial Intelligence?

People should be able to identify AI-generated material, AI-assisted work, AI interactions and AI-supported decision-making.

Labels should describe AI involvement accurately. A label is a disclosure mechanism, not automatically a guarantee of correctness, security or authenticity.

AI GENERATED

Artificially generated material

Indicates that AI generated substantial content such as text, images, audio or video.

AI ASSISTED

AI-assisted creation

Indicates meaningful AI assistance in otherwise human-led production.

AI INTERACTION

Interacting with AI

Discloses that the user is communicating with an AI-driven interface or service.

AI DECISION SUPPORT

AI-supported decisions

Indicates that AI contributes to a recommendation, assessment or decision.

A machine-readable disclosure proposal

Human-readable labels could be accompanied by structured metadata describing AI involvement. This is an illustrative proposal, not an approved international standard.

{
  "schema": "https://aig.org/labelling/",
  "version": "0.1-proposal",
  "aiDisclosure": {
    "category": "AI_GENERATED",
    "declaredBy": "Example Publisher",
    "provider": "Example AI Provider",
    "model": "Example Model",
    "humanReviewed": true,
    "declaredAt": "2026-11-06",
    "description": "Illustrative AI disclosure"
  }
}

Design objectives

Clear and unambiguous public disclosure.
Consistent human-readable labels.
Machine-readable representation.
Provider and publisher identification.
Disclosure of meaningful human involvement.
Support for verification and provenance.
Compatibility with relevant existing standards.
Accessible presentation across devices.

Proposal status

These concepts are discussion proposals. They are not certification marks and do not claim formal regulatory or standards-body approval.