Playbook Human-Made Content Standard

Last updated: 23 August 2026

1. The Playbook Principle

Playbook is an archive for human-made creative and cultural work.

The purpose of this Standard is to protect that principle while recognizing that modern creative tools range from traditional physical media to digital cameras, editing software, procedural systems, 3D software, and computational tools.

Playbook does not require creators to avoid technology.

Playbook requires that the creative work archived on the platform remain meaningfully created by people rather than generated by artificial-intelligence systems.

2. The Core Rule

Content presented on Playbook as creative work must be human-made.

You must not submit creative material that was wholly or materially generated by a generative artificial-intelligence system.

This includes content produced from:

• text prompts;
• image prompts;
• voice prompts;
• video prompts;
• generated latent imagery;
• generative image-to-image systems;
• generative video systems;
• generative audio or music systems;
• generative 3D systems;
• generative text systems where the generated text itself forms part of the submitted creative work.

The use of ordinary digital tools does not automatically make work non-human.

The question is whether a person created the expressive substance of the work or whether a generative system materially synthesized it.

3. What “Human-Made” Means

For purposes of Playbook, work is human-made when its meaningful creative expression originates from human decisions, human execution, or human-directed non-generative processes.

Examples may include:

• drawing;
• painting;
• sculpture;
• photography;
• typography;
• graphic design;
• filmmaking;
• animation created by human artists;
• physical craft;
• architecture;
• industrial design;
• analogue processes;
• digital illustration;
• manually modeled 3D work;
• manually composed music;
• human-written creative text;
• human-designed interactive work;
• procedural work where the creator designed the rules, parameters, system, and final selection without generative AI creating the expressive content.

Human-made work may still involve software.

Software assistance alone does not transfer authorship from a person to a machine.

4. Generative AI Is Not Permitted

The following may not be submitted as human-made creative work on Playbook:

• text-to-image output;
• text-to-video output;
• text-to-music or generative audio output;
• image-to-image generative output;
• AI-generated illustrations;
• AI-generated photographs;
• AI-generated paintings;
• AI-generated designs;
• AI-generated architecture concepts;
• AI-generated film shots;
• AI-generated animation;
• AI-generated textures where those textures materially form part of the submitted work;
• AI-generated characters;
• AI-generated faces;
• AI-generated backgrounds;
• AI-generated objects;
• AI-generated visual effects;
• AI-generated 3D assets;
• synthetic voice performance presented as human performance;
• AI-generated music or substantial musical passages;
• AI-generated substantive writing presented as human-authored creative or archival writing.

Changing, retouching, filtering, tracing, repainting, compositing, or otherwise modifying generated material does not automatically make the resulting work human-made.

If generative AI created a material part of the expressive content, the work remains ineligible under this Standard.

5. Mixed Human + AI Work

A work is not eligible merely because a human contributed some portion of it.

If generative AI materially created part of the final expressive work, the submission is generally not permitted.

Examples include:

• a human drawing placed over an AI-generated background;
• an AI-generated character inserted into a human photograph;
• AI-generated texture used as a major design element;
• AI-generated architecture imagery edited by a human;
• an AI-generated photograph heavily color-graded afterward;
• AI-generated music combined with human vocals;
• AI-generated frames inserted into human-made film or animation.

Playbook's rule concerns the final archived work, not the percentage of human labor.

There is therefore no “mostly human” threshold that automatically permits generative content.

6. Generative Fill and Generative Expansion

Generative Fill, generative expansion, outpainting, and similar tools are not permitted when they synthesize new visual content that appears in the final submitted work.

Examples include:

• extending a landscape beyond the photographed frame;
• generating missing portions of a subject;
• replacing a removed object with synthesized scenery;
• creating new clothing, hair, faces, objects, buildings, or backgrounds;
• generating additional elements within an existing artwork.

Traditional manual compositing, cloning, painting, cropping, or retouching remains distinguishable from prompt-based or generative synthesis.

7. Object Removal and Retouching

Removing imperfections or unwanted elements does not automatically violate this Standard.

Generally permitted techniques include:

• cropping;
• manual clone stamping;
• manual healing;
• dust removal;
• blemish correction;
• exposure adjustment;
• contrast adjustment;
• color correction;
• color grading;
• dodging and burning;
• sharpening;
• ordinary image compositing using human-created source material;
• manual restoration.

However, if the removal process uses a generative model to synthesize material new content into the resulting image, it may fall under the prohibition on generative fill.

Creators should use non-generative techniques where there is uncertainty.

8. Computational Photography

Photographs are not disqualified merely because the camera or phone uses computational processing.

Ordinary camera functions may include:

• autofocus;
• auto-exposure;
• white-balance adjustment;
• HDR merging;
• noise reduction;
• image stabilization;
• lens correction;
• demosaicing;
• sensor processing;
• computational sharpening;
• subject detection used for camera operation.

These processes generally do not violate this Standard where they enhance or process a photograph captured by the creator rather than generating materially new scenes, subjects, objects, or expressive content.

A feature marketed as “AI” is therefore not automatically prohibited.

Its actual function matters.

9. AI-Assisted Editing

Playbook distinguishes assistance from generation.

Tools may be permitted where they analyze, select, organize, correct, or enhance human-created material without materially synthesizing new expressive content.

Examples may include:

• automatic masking;
• subject selection;
• sky selection used only to create a mask;
• edge detection;
• automatic exposure suggestions;
• automatic color correction;
• stabilization;
• non-generative denoising;
• spell checking;
• grammar checking;
• metadata organization;
• search or asset organization.

The same tool may include both acceptable and prohibited features.

For example, automatic subject selection may be acceptable while generative replacement of the selected subject would not be.

Creators are responsible for how they use the tool.

10. Upscaling and Denoising

Conventional enlargement, interpolation, sharpening, and non-generative denoising are generally permitted.

An enhancement tool becomes problematic where it materially invents visual detail that was not reasonably present in the source.

Generative super-resolution or enhancement systems that create new facial features, textures, objects, lettering, or other expressive detail may make a work ineligible.

Where the difference between enhancement and generation is unclear, creators should avoid using the tool on work intended for Playbook or retain evidence describing the workflow.

11. Background Removal

Removing a background from human-created material is generally permitted.

Replacing that background with:

• another human-made photograph;
• a human-created illustration;
• manually designed graphic material;

may also be permitted where the resulting work otherwise complies with Playbook's rules.

Replacing the background using generative AI is not permitted.

12. Manual Compositing

Digital compositing remains human-made when the creator manually combines eligible human-made source material.

Examples include:

• collage;
• photomontage;
• compositing multiple photographs;
• combining hand-drawn and photographed elements;
• combining manually created 3D renders and photography.

Every materially expressive source element used in the final work should itself comply with this Standard.

An AI-generated source asset does not become eligible merely because it is incorporated into a larger human-made composition.

13. 3D Work

Human-created 3D work is permitted.

This may include:

• manual modeling;
• sculpting;
• rigging;
• lighting;
• texturing using human-created or otherwise eligible source assets;
• rendering;
• simulation;
• manually designed procedural systems.

Use of a 3D application does not make work synthetic under this Standard.

However, generative systems that create substantial models, scenes, characters, textures, animation, or other expressive assets from prompts may make the final work ineligible.

14. Procedural and Generative Art Systems

Traditional procedural and algorithmic art is not automatically the same as modern generative AI.

A work may remain eligible where the human creator meaningfully designs:

• the rules;
• algorithm;
• parameters;
• geometry;
• simulation;
• system behavior;
• inputs;
• composition;
• selection of the final result.

Examples may include:

• mathematical art;
• shader art;
• particle systems;
• rule-based visual systems;
• parametric architecture;
• procedural geometry;
• manually programmed interactive artwork.

Playbook may distinguish these processes from systems where a trained generative model synthesizes expressive content from prompts.

When classification is uncertain, Playbook may request information about the workflow.

15. Photography of Human-Made Objects and Works

Playbook may archive photographs documenting human-made objects and creative works.

For example:

• automobiles;
• architecture;
• watches;
• printed ephemera;
• paintings;
• sculpture;
• technology;
• historical design objects.

The photograph and the work being documented may involve different creators.

Users should identify the original maker or creator accurately where known.

The fact that a user photographed an object does not mean the user created the underlying object.

16. Archival and Historical Reproductions

Digitization does not automatically make material synthetic.

Examples that may remain eligible include:

• scans of photographs;
• scans of printed material;
• film scans;
• restoration of damaged photographs;
• digitization of analogue recordings;
• documentation of physical artifacts.

Restoration should not introduce generative content that materially invents missing creative detail.

Playbook may permit careful reconstruction or restoration where the process is substantially human-directed and transparently archival rather than generative fabrication.

17. Creative Text and Archive Context

Playbook's human-made principle also applies to substantive creative or archival text submitted as your own writing.

Users should not present substantial generative-AI output as their own human-written:

• Story;
• Cultural Shift commentary;
• creative writing;
• critical interpretation;
• original archival description.

Tools that assist with:

• spelling;
• grammar;
• punctuation;
• formatting;
• basic translation;
• accessibility;

do not automatically make text ineligible where the user remains the substantive author.

Factual information obtained from research tools should be independently reviewed before publication.

Users remain responsible for the accuracy of what they submit.

18. Research and Inspiration

Using technology during research does not by itself determine whether the final work is eligible.

The Human-Made Content Standard primarily concerns the actual material submitted to Playbook and the process used to create it.

A creator may use ordinary reference material, research databases, books, archives, websites, or software tools while working.

However, incorporating generative AI output materially into the final work makes the work ineligible under this Standard.

19. Human-Made Confirmation

Before publishing, Playbook may require a user to confirm:

“I confirm that this submission complies with Playbook's Human-Made Content Standard and does not contain prohibited generative or synthetic creative content.”

Playbook may record:

• the user's account ID;
• the Human-Made Content Standard version;
• the confirmation;
• the server-recorded time;
• the related contribution;
• relevant moderation results.

A deliberately false confirmation is a violation of Playbook's rules.

20. Verification and Moderation

Playbook may use several methods to evaluate whether submitted work complies with this Standard.

These may include:

• automated AI-content classifiers;
• file metadata;
• provenance metadata;
• Content Credentials or similar provenance information;
• upload history;
• comparison of source material;
• moderation evidence;
• human review;
• information provided by the user.

No single technical signal is necessarily conclusive.

Automated classifiers can produce false positives and false negatives.

Metadata may be missing, removed, altered, or incomplete.

The absence of provenance metadata does not automatically mean that content was AI-generated.

The presence of provenance metadata does not automatically prove that all claims about a work are true.

Playbook should consider the available evidence as a whole.

21. Content Credentials and Provenance

Where available, Playbook may consider provenance technologies such as Content Credentials as one source of information about a file's creation or editing history.

Creators are not required to possess Content Credentials unless Playbook announces a future requirement for a particular feature.

A creator should not be penalized merely because:

• a camera does not support provenance credentials;
• software removed metadata;
• an older artwork has no digital provenance record;
• an analogue work was digitized.

Playbook's purpose is to strengthen provenance over time without excluding legitimate human-made work that predates modern provenance technologies.

22. Requests for Workflow Evidence

Where a submission raises a significant authenticity question, Playbook may request reasonable information about how the work was created.

This may include:

• source photographs;
• sketches;
• working files;
• layer history;
• export history;
• RAW files;
• project files;
• screenshots of the creative workflow;
• provenance metadata;
• other information reasonably capable of clarifying the process.

Playbook should request only information reasonably relevant to the review.

Failure to possess a particular form of evidence does not automatically prove that a work is AI-generated.

23. False Positives

Playbook recognizes that AI-detection systems can incorrectly flag legitimate human-made work.

A moderation signal is not an accusation of dishonesty.

Where Playbook believes further review is appropriate, the work may be held privately while the question is evaluated.

Creators should have a reasonable opportunity to request reconsideration.

24. Human Review

Human reviewers may consider:

• the submitted work;
• automated scores;
• source material;
• workflow information;
• creator explanation;
• metadata;
• provenance records;
• previous related work;
• other relevant evidence.

Human review should focus on whether the work complies with this Standard rather than whether a reviewer personally likes the work or its style.

25. Review Requests

If you believe Playbook incorrectly rejected or removed human-made work, contact:

hello@playbook.cam

Subject:

Human-Made Review

Include:

• the affected work;
• your Playbook account information;
• a brief explanation of the process used to create the work;
• any supporting information you believe is relevant.

Do not send sensitive personal information that is unrelated to the review.

26. Deliberate Misrepresentation

Playbook may take stronger action where there is evidence that a user knowingly:

• submitted prohibited generated content;
• falsely attested that generated content was human-made;
• altered evidence to conceal generative AI use;
• repeatedly attempted to evade detection;
• created alternate accounts to continue the same conduct.

Depending on severity and history, enforcement may include:

• rejection;
• removal;
• publishing restrictions;
• account restrictions;
• suspension;
• termination.

An incorrect automated flag by itself is not evidence of deliberate misrepresentation.

27. Existing Work and Historical Collections

The Human-Made Content Standard applies to content submitted to Playbook regardless of when the underlying work was created.

Historical human-made creative work is welcome.

Older work is not required to have modern creation metadata or digital provenance.

Playbook may rely more heavily on contextual and human review for older material.

28. Future Technologies

Creative technology will continue to change.

Playbook will evaluate new tools according to the underlying principle of this Standard:

Did the human creator make the expressive work, or did a generative system materially synthesize it?

Product names and marketing terminology are not decisive.

A feature called “AI” may perform a non-generative editing task.

A feature that avoids the word “AI” may still generate synthetic content.

Playbook evaluates the function rather than the label.

29. Updates to This Standard

Playbook may update this Standard as creative technologies and moderation methods evolve.

Each published version will have:

• a version number;
• an effective date.

Users may be required to accept a new version before submitting additional work.

Playbook may retain historical acceptance records showing which version applied when a contribution was submitted.

Changes should not retroactively convert legitimately accepted human-made work into misconduct merely because terminology or technical guidance later changes, unless continuing to host that content creates a legal or serious safety issue.

30. The Promise Playbook Can Actually Make

Playbook's promise is that generative and synthetic creative content is prohibited under the platform's rules and that Playbook will build systems intended to enforce that principle.

Playbook does not claim that any automated detector is perfectly accurate.

Playbook does not claim that prohibited material can never evade detection.

Playbook therefore avoids describing the platform as technologically guaranteed to contain “zero AI.”

The standard is:

Human-made work is the rule. Generative creative content is not permitted. Uncertain cases can be reviewed by people.

31. Contact

Questions about this Standard may be sent to:

hello@playbook.cam

Subject:

Human-Made Standard

End of Playbook Human-Made Content Standard — Version 1.0