STANDARD V3.84CHARITY-GOVERNED

THE CIP STANDARD

The framework for subsisting AI⇆IP rights for creative intellectual property. Developed by Creative Intellectual Property Charity with input from creators, agencies, lawyers, underwriters, and platform operators.

View certification →Read the glossary

Standard at a glance

Version

3.84

Purpose

Subsisting AI⇆IP rights framework

Applies to

Creators, agencies, legal, underwriters, platforms

Core output

cip.md declaration file

Status

Adopted by Trustee Board

Published November 2025

Revised April 2026

Review cycle 6-monthly

Licence All Rights Reserved

Reading 18 min

Abstract

The CIP Standard sets out a clear framework for creative content provenance, rights awareness, consent, governance, and professional recognition. It exists because creative works carry subsisting IP rights that persist across every form of AI use, from training data ingestion to synthetic content generation to automated distribution. The standard provides a shared structure for creators, organisations, advisers, and platforms working to understand and professionally manage those rights.

Contents

01Preliminary: the renaming problem 02Subsisting rights and registered rights 03The six functional areas 04Machine-readable declaration: cip.md 05How the standard is organised 06Governance, oversight, and versioning

01

01

Preliminary: the renaming problem

The core problem is definitional. When a creative work is renamed “training data”, the rights that subsist in it do not disappear. Copyright does not lapse. Neighbouring rights are not waived. Identity rights are not surrendered. Yet the operational frameworks governing AI training pipelines have treated those rights as if they had.

CIP exists to correct that: to bring structure, professional standards, and recognised evidence to the question of what rights subsist, in what content, and on what terms AI use is or is not permitted.

A common standard helps create consistency. It gives creators, organisations, advisers, and platforms a shared basis for understanding what is expected, what evidence matters, and what recognised good practice looks like.

1.1

The term “training data” is an industry category designed to describe the inputs to machine learning systems. It is not a legal category and does not change the legal status of the content it describes.

02

02

Subsisting rights and registered rights

This distinction is fundamental to understanding why the CIP framework exists and what problem it solves.

Subsisting rights

Automatic legal recognition

A subsisting right exists the moment a qualifying work is created or a qualifying act occurs. No application. No registration. No fee. No government office involved. The right comes into existence automatically under law.

Registered rights

State-issued exclusivity

A registered right is granted by a state authority following a formal application and examination process. Registration requires active steps by the rights holder and, typically, payment of fees. Examples include patents, trade marks, and registered designs.

Definitions: subsisting rights

Copyright Subsists from creation in songs, photographs, writing, and other original works. No application required.

Moral rights Subsist alongside copyright, protecting the author’s right to attribution and the integrity of their work.

Neighbouring rights Subsist in a performance or recording from the moment it is fixed. Protect performers and producers.

NILP rights Name, image, likeness, and persona rights subsist as commercial identity markers. The fastest-moving area of AI law.

Database rights Subsist in structured datasets where sufficient investment was made in their creation.

Biometric rights Subsist in faceprints, voiceprints, and other biological identifiers used by AI systems.

Why this distinction matters for AI

The AI content pipeline almost exclusively engages subsisting rights, not registered rights. When an AI company scrapes the web, ingests a music catalogue, or trains on photographs, the rights it encounters are automatic and universal. They do not depend on whether any creator filed a patent or registered a trade mark.

This is why renaming creative content as “training data” does not help AI companies legally. The copyright subsists. The neighbouring rights subsist. The NILP rights subsist. None of them required registration. None of them lapse because the work was ingested.

2.1

The standard IP categorisation distinguishes between rights that arise automatically (subsisting) and those that require formal state grant (registered). CIP is built around subsisting rights because they are what AI pipelines encounter universally.

03

03

The six functional areas

The CIP Standard brings together core areas that increasingly shape the handling of creative content. It is organised around distinct functions in the creative rights chain.

Provenance Where content comes from and how it is evidenced.AI⇆IP relevance: Establishing chain of origin before AI ingestion.

Rights awareness The rights that may exist in a work, asset, performance, likeness, or dataset.AI⇆IP relevance: Identifying subsisting rights before and after AI use.

Consent How permission is granted, limited, reviewed, or withdrawn.AI⇆IP relevance: Consent to AI training; temporal and scope-limited licences.

AI pipeline exposure How content moves through AI training, generation, and distribution stages.AI⇆IP relevance: The specific pipeline stages where subsisting rights are engaged.

Governance Responsibilities, controls, and oversight across participants.AI⇆IP relevance: Accountability for rights-aware ingestion and output.

Recognition Credentials, marks, and formal recognition against defined standards.AI⇆IP relevance: Verified trust signals for rights-compliant AI operations.

3.1

The six functional areas map to the five certification tracks: Creator, Agency, Legal, Underwriter, and Platform. Each track engages with all six areas from the perspective of its specific professional function.

04

04

Machine-readable declaration: cip.md

The cip.md file is the machine-readable expression of the CIP Standard. It sits alongside creative content, in the same directory, attached to the same asset, deployed at the same endpoint, and declares the rights, consent terms, and licensing conditions that apply to the work.

The format is designed to be readable by both humans and AI systems. It uses a structured key-value syntax that can be parsed programmatically while remaining intelligible to anyone who opens the file. The parallel with robots.txt is deliberate: a lightweight, file-based declaration that AI systems are expected to read and respect.

CDR

A Core Data Record makes subsisting rights visible and evidenced.

cip.md

A cip.md declaration makes those rights machine-readable.

Registry

The Rights Registry makes them independently verifiable.

4.1

The cip.md format follows the convention established by robots.txt: a plain-text file placed at a known location that automated systems are expected to read and respect. The full specification is available on the cip.md specification page. What is a markdown file?

05

05

How the standard is organised

The CIP Standard is organised through role-based certification paths. Each path maps to a distinct function in the rights chain and has its own requirements, assessments, standards, and outputs.

Creator Certification Individuals who originate rights-bearing content.

Agency Certification Organisations managing rights across client portfolios.

Legal Practitioner Certification Legal professionals advising on rights and regulation.

Underwriter Certification Underwriters and risk professionals.

Platform Certification Platforms, distributors, broadcasters, and brands.

5.1

Each pathway is separate. Recognition in one path does not replace the requirements of another.

06

06

Governance, oversight, and versioning

Governance

CIP is governed by Creative Intellectual Property Charity through a standards structure with representation from the professional communities it serves. Oversight includes standards governance, independent audit support, and ongoing review as regulation and professional practice develop.

Versioning

The CIP Standard is currently at version 3.84. As legal frameworks, professional practice, and sector expectations develop, particularly around AI training obligations, synthetic content disclosure, and identity rights, the standard will be revised, expanded, and updated. Versioning ensures changes are structured, visible, and reviewable.

6.1

This document is reviewed on a 6-monthly cycle by the Standards Committee. The next scheduled review is October 2026. Annotations in this document follow the CIP convention: marginal notes provide context; they do not form part of the normative standard text.

Notes on this document

Consultation responses and feedback on the standard should be directed to charity@creativeip.org.

Governance body: Creative Intellectual Property Charity (CIO Foundation Model, application 5289637).

Related documents

CIP Level 1 Foundation Course

7 modules Certification Tracks 5 professional paths CIP Glossary 389 defined terms AI IP Risk Self Assessment 8 questions

End of document

CIP Standard v3.84

THE STANDARD IS PUBLISHED.

The courses are open. The tools are available. Start with the self assessment to understand your rights exposure, or go straight to the Level 1 course.

Take the self assessment →Begin Level 1 course