Creative Disclosure Standard

Clearer origins. A more informed world.

01 / Purpose

What is the CDS?

The Creative Disclosure Standard (CDS) is a voluntary, creator-driven classification system that identifies how a work was created and what tools were used in its production. The goal is simple: greater clarity for readers, respect for creators, and a more informed creative landscape.

The CDS is not a judgment of quality. It is a disclosure of process. By sharing how a work was created, we empower readers to make informed choices and support a creative ecosystem built on transparency, respect, and trust.

02 / Classification

The CDS categories

Examples

  • A novel handwritten entirely on paper and edited manually.
  • A painting created with physical brushes, pigments, and canvas.
  • A musical composition written by hand and performed on acoustic instruments without digital production assistance.
  • A traditionally sculpted work created using manual tools.

Classification boundary

A work does not qualify as 100% Human / Analog merely because no generative AI was used. If conventional digital equipment or software materially participates in the production process, the appropriate classification is generally Machine-Assisted.

Suggested disclosure

“This work was created entirely through human effort using traditional, analog methods.”

Examples

  • Writing a novel using a conventional word processor.
  • Creating illustrations using Photoshop or Illustrator without generative AI features.
  • Using conventional spelling and grammar correction.
  • Producing music using digital recording software and conventional audio-processing effects.
  • Editing photographs using traditional exposure, color, cropping, and retouching tools.
  • Creating digital animation through manually controlled animation software.

Classification boundary

The mere presence of sophisticated software does not automatically establish AI participation. The relevant distinction is what the software actually performs. Conventional automation that executes human-directed operations without generating new expressive material is Machine-Assisted. When generative AI makes a substantive creative contribution, another classification may apply.

Suggested disclosure

“This work was created by a human author or artist using conventional digital production tools without generative AI contributions to its expressive content.”

Examples

  • Using AI to identify spelling, punctuation, or grammatical errors.
  • Using AI to suggest alternative wording for human-written sentences.
  • Requesting editorial feedback on pacing, clarity, or consistency.
  • Using AI to identify continuity problems in a manuscript.
  • Receiving structural suggestions for an existing human-authored work.
  • Using AI to identify or select suitable entries from a human-created dialogue thesaurus.
  • Using AI to help organize research, outlines, or existing creative material.

Classification boundary

AI-Assisted describes a supporting role. AI may propose local revisions or limited expressive language, but its contribution does not amount to substantial independent origination of the completed work. When AI transforms a substantial preexisting human-authored work into a different expressive form, AI-Adapted may be more appropriate. When both the human and AI originate substantial portions of the work's expressive content, AI-Co-Created may apply.

Suggested disclosure

“This work was created primarily through human authorship with artificial-intelligence assistance for editing, refinement, organization, or other supporting tasks.”

Examples

  • Converting an original human-written screenplay into prose using AI.
  • Transforming a human-written stage play into a narrative adaptation.
  • Converting human-created storyboards into animated sequences using generative AI.
  • Turning a substantial human-authored narrative into another expressive format through AI-assisted transformation.
  • Using generative AI to reexpress an existing human-created work while maintaining its foundational creative decisions.

Classification boundary

AI-Adapted is not simply another name for AI-Generated. A completed adaptation may contain substantial AI-produced language, images, or other expression while remaining rooted in a preexisting human-created work. The classification recognizes both the human provenance of the source material and the transformative role of artificial intelligence. A prompt, brief concept, character name, or short instruction does not by itself constitute a substantive source work. If AI independently originates the underlying creative material, AI-Co-Created or AI-Generated may be more appropriate.

Suggested disclosure

“This work was adapted using generative artificial intelligence from an original human-created source. The underlying creative material originated with the human creator, while AI assisted in transforming it into its final expressive form.”

Examples

  • A human writer and generative AI alternately develop substantial original portions of a story.
  • An artist combines AI-generated imagery with substantial original illustration, graphic design, and compositional work.
  • A designer uses AI-generated source imagery and independently constructs the final composition, typography, layout, and visual presentation.
  • A musician combines substantial human-composed passages with AI-generated musical material.
  • A filmmaker combines human-created scenes and substantial generative AI sequences into a unified work.

Classification boundary

AI-Co-Created requires more than human prompting, approval, or superficial editing. There must be a meaningful human creative contribution to the resulting expression, beyond merely instructing AI to produce the work. Likewise, AI participation must be substantive rather than incidental. The classification does not imply that an AI system is a legal author, rights holder, or person. The term co-created describes participation in the creative process, not legal authorship.

Suggested disclosure

“This work was co-created through substantial human creative contributions and generative artificial intelligence. Both contributed expressive material to the final work.”

Examples

  • A complete novel drafted primarily by generative AI from human prompts, with human proofreading afterward.
  • A collection of illustrations generated almost entirely by AI, with minimal human modifications.
  • An AI-produced musical composition based primarily on text instructions.
  • A video whose scenes and principal visual content are overwhelmingly generated by AI.
  • An AI-produced article or narrative receiving only minor human editorial revisions.

Classification boundary

The CDS threshold for AI-Generated is 75% of the completed work's substantive expressive content. This is a self-assessed classification threshold, not a claim that software can reliably measure AI involvement. The proportion should be evaluated according to the nature of the medium and the content actually originated by AI, rather than automatically equating word counts, file sizes, working hours, or numbers of prompts with creative contribution. An important exception in classification is identifiable preexisting human-created source material. Where AI substantially transforms such a source, AI-Adapted may more accurately describe the process than AI-Generated, even if the adaptation contains extensive AI-produced expression.

Suggested disclosure

“This work was primarily generated using artificial intelligence, with human direction, selection, review, and editorial oversight.”

03 / Guidelines

Disclosure Guidelines

Scope of Classification: The Finished Creative Work

The Creative Disclosure Standard (CDS) classifies the creative content of a finished work, not the preliminary activities undertaken during its development.

Research, brainstorming, idea exploration, conceptual discussions, and other preparatory activities fall outside the scope of CDS classification, regardless of whether artificial intelligence was used.

The use of AI during these activities does not, by itself, determine or alter a work's classification.

CDS classification is based on the origin and nature of the expressive content incorporated into the finished creative product.

Ideas and information developed through AI-assisted research or brainstorming do not constitute AI-generated content unless AI-generated expressive material is incorporated into the completed work.

The governing principle: CDS classifies what was produced, not how the creator arrived at the idea.

How CDS Classifications Are Determined

Classify the Creative Process, Not Merely the Software

The use of a particular software application does not automatically determine the classification. A single application may offer conventional editing tools, machine-learning utilities, and generative AI features. CDS classification depends on what the technology contributed to the final work. For example, conventional image color correction differs from generating an entirely new image. Correcting punctuation differs from generating an original chapter. The appropriate classification reflects the actual contribution, not merely the name or marketing description of the tool.

Distinguish Assistance, Adaptation, and Origination

Three questions help identify the appropriate classification:

  1. What did the human creator originally produce? Determine whether the human created the underlying expressive work, a substantial source artifact, individual elements, or only prompts and instructions.
  2. What did the technology contribute? Identify whether it performed conventional processing, editorial assistance, transformation of existing material, or original generation of expressive content.
  3. How was the completed work constructed? Consider the substantive creative contributions present in the final work, rather than relying solely on how much time a person spent directing the software.

These questions distinguish supportive tools from transformative tools and generative production.

AI Use Does Not Automatically Determine the Classification of an Entire Work

A creative project may involve several types of technological assistance. For example, a human-written novel may use AI for proofreading while its cover uses AI-generated source imagery. The manuscript and cover involve different creative processes and may appropriately carry different classifications. Similarly, using AI for one limited editorial task does not necessarily turn an entire work into an AI-Assisted production when that contribution is incidental and does not materially affect the completed expression. The classification should describe the substantive production method of the work or component being disclosed. Creators may voluntarily provide separate disclosures for distinct components when this better communicates the production process.

The 75% AI-Generated Threshold

CDS establishes a 75% substantive expressive-content threshold for the AI-Generated classification. The threshold is intended to distinguish works predominantly originated by generative AI from works reflecting substantial human origination or adaptation. It is not a scientific measurement of creativity.

For text, relevant considerations may include who originated the narrative, actual written expression, and substantive portions of the completed manuscript. For visual art, relevant considerations may include the origination of principal visual elements and the human contribution to the finished composition. For audio and audiovisual media, considerations may include the origination of the principal expressive elements, such as music, performance, narration, dialogue, and visual sequences.

The threshold is applied through honest creator self-assessment. Falling below 75% AI-generated content does not automatically make a work AI-Assisted or AI-Co-Created. The appropriate classification still depends on whether the remaining process involved adaptation from an existing human source or substantive shared origination.

Provenance Statements

A CDS classification identifies the broad creative-production method. A provenance statement provides additional information describing how the work was produced. Provenance statements are optional under the baseline CDS framework. They may identify:

  • the human-created source material used in an AI adaptation;
  • the specific roles performed by generative AI;
  • which components were created by humans and which incorporated AI-generated material;
  • the nature of the creator's original contributions;
  • the creator's editorial and production involvement.

Example — AI-Adapted novel: 'Based on an original human-written screenplay. Generative AI was used to adapt the screenplay into narrative prose while preserving the creator's underlying story, characters, dialogue, and creative direction. The completed adaptation was reviewed and edited by the creator.'

Example — AI-Co-Created book cover: 'AI-generated source imagery was incorporated into an original human-created design. Final composition, typography, arrangement, and visual presentation were completed by the designer using conventional graphics software.'

Example — AI-Assisted manuscript: 'The manuscript was written by the author. Artificial intelligence was used for editorial feedback, language refinement, and continuity review.'

These statements provide additional clarity without imposing a mandatory documentation burden.

Where Should CDS Disclosures Appear?

For books, a CDS classification may appear in the copyright or front-matter pages. Creators may also include the classification in publication metadata, product descriptions, or other suitable locations. For visual, audio, and audiovisual works, disclosures may appear in credits, accompanying descriptions, packaging, or relevant production information.

CDS encourages disclosures that are clear, accessible, and understandable to the intended audience. A front-matter disclosure is sufficient for a book under the baseline CDS framework. Additional provenance information is optional.

The CDS Principle

Transparency without judgment. The Creative Disclosure Standard recognizes that creative expression evolves alongside the tools used to produce it. Traditional craftsmanship, digital production, AI assistance, AI adaptation, human-AI collaborative production, and primarily AI-generated works represent different creative processes. CDS provides a common language for describing those differences.

The decision to value, purchase, publish, support, or reject a work remains with the individual. CDS tells audiences how a work was made. It does not tell them what to think of it.

04 / Governance

Administration & Classification

The Creative Disclosure Standard establishes a common framework for identifying and communicating the methods used in the creation and production of published works. CDS defines the classification categories, their general meanings, and the disclosure marks associated with each category.

Classification Responsibility

The responsibility for evaluating and assigning a CDS classification rests with the creator, publisher, or authorized organization implementing the standard. Participating organizations may establish their own internal procedures for evaluating creative workflows, determining appropriate classifications, and resolving borderline cases, provided their decisions remain consistent with the published CDS category definitions. CDS does not require individual works to undergo independent review, certification, or approval. Organizations should be prepared to disclose which party assigned a given classification.

Industry Implementation

Publishers, ISBN registration agencies, distribution platforms, and other participating organizations may incorporate CDS classifications into their existing publication and metadata workflows. Organizations may develop implementation guidelines appropriate to their publishing operations. Implementation must not redefine the published category meanings or introduce additional CDS categories.

Purpose and Limitations

CDS identifies the nature of human and machine involvement in creative production. It does not determine copyright eligibility, establish creative ownership, evaluate artistic merit, or certify the accuracy of individual disclosures. A CDS classification is a statement of creative provenance, not an endorsement or certification by the Creative Disclosure Standard.

05 / Example

In use

A CDS disclosure shown on a novel’s copyright page, where readers can see both the selected category and its definition. This example is classified AI-Adapted.

Open novel copyright page showing an AI-Adapted Creative Disclosure Standard badge and the six CDS categories
06 / Use

Usage terms

  1. 01

    One badge per work — pick the tier matching the highest level of AI involvement in its creation.

  2. 02

    Display it unaltered — no recoloring, stretching, or rewriting the definitions.

  3. 03

    Put it where people see it first — cover, product page, credits, description. A disclosure nobody sees isn't one.

  4. 04

    Tell the truth — the badge is a self-declaration. Mislabeling guts the whole standard.

  5. 05

    No certification implied — CDS describes origin, not quality. It doesn't endorse the work.

  6. 06

    Free to use, no license — anyone can display the badges under these terms, for now.

07 / Downloads

Get the badges

Choose the tier that matches your process, then download the artwork or copy an embed for your website, README, or newsletter.

CDS identity assets Stacked logo Horizontal logo Circular seal
08 / Tier picker

Which badge fits your work?

09 / CDS v0.1

Machine-readable spec

For websites, ebooks, and tools: declare your tier in code, not just pixels. Vocabulary: https://creativedisclosure.org/vocab/1.0

HTML meta tags

<meta name="cds-tier" content="ai-adapted">
<meta name="cds-vocab" content="https://creativedisclosure.org/vocab/1.0">

Replace ai-adapted with your own tier slug.

JSON-LD

{
  "@context": ["https://schema.org", {"cds": "https://creativedisclosure.org/vocab/1.0#"}],
  "@type": "CreativeWork",
  "cds:tier": "https://creativedisclosure.org/tiers/ai-adapted"
}

Visible badge embed

<a href="https://creativedisclosure.org/tiers/ai-adapted"><img src="https://creativedisclosure.org/badges/ai-adapted.svg" alt="Creative Disclosure Standard: AI-Adapted — This work originated from a substantial human-created source and was adapted using generative AI tools." width="120"></a>

Reserved for v0.2

EPUB OPF metadata, image XMP/EXIF fields, and audio ID3/Vorbis comment mappings are documented for a future version, not implemented in v0.1.

Canonical slugs

human-analog, machine-assisted, ai-assisted, ai-adapted, ai-co-created, ai-generated

Tier URIs follow the pattern https://creativedisclosure.org/tiers/<slug>.

10 / Registry

Registry

Works carrying the CDS badge.

A public record of works that disclose their AI involvement with Creative Disclosure Standard badges. The founding entries are the creators' own catalog — fifteen volumes across three arcs, all badged AI-Adapted.

Submit your work

Badging your work with CDS? Public submissions open soon.

11 / Questions

Questions

Where is the line? Do I disclose my word processor? My spellchecker?

CDS draws the line at generative participation. Typewriters, word processors, spellcheckers, grammar tools: these carry or correct your expression. That is Machine-Assisted. The AI tiers start where a system generates or transforms creative expression. It drafts prose. It invents scenes. It restructures the work. Your spellchecker fixes your sentences. Generative AI writes new ones. If the tool shaped the creative content, disclose it. If it only handled the typing, you don't have to.

Why categories instead of a 0–100 score?

Categories describe the process that produced the work — something a publisher can stand behind and a reader can understand. AI-Adapted means a human-authored source work exists and AI transformed its expression. That claim is checkable; '37% AI' is not a measurement anyone can take. The 75% threshold for AI-Generated is applied through honest creator self-assessment, not precise measurement: it marks the point where a work is predominantly AI-originated.