The Provenance Protocol: Chain of Intellectual Custody from First Note to Final Publication

The Provenance Protocol: Chain of Intellectual Custody from First Note to Final Publication

DOI: Pending assignment

John Swygert

August 1, 2026

Abstract

Modern authorship is increasingly collaborative, tool-assisted, distributed, and difficult to reconstruct after publication. A book may involve an originating author, artificial intelligence systems, editors, ghostwriters, researchers, translators, designers, software, and publishers. Existing disclosure practices rarely preserve the full developmental history of the work.

This paper proposes the Provenance Protocol, a structured chain of intellectual custody beginning before drafting and continuing through publication. The protocol does not prescribe one acceptable route to authorship. It permits solo, collaborative, ghostwritten, AI-assisted, edited, translated, and mixed workflows. Its purpose is to make the chosen route explicit, updateable, verifiable, and accountable.

The protocol begins with an Authorship Process Declaration describing the intended creative method and anticipated contributors. It then preserves the originating note, outline, outline development, source record, contribution log, drafts, revisions, approvals, and publication record. Changes in method are documented rather than prohibited.

The paper distinguishes three levels of disclosure: a concise public statement, a more detailed verifiable project record, and a private evidentiary archive. It proposes metadata fields, version controls, contributor classifications, final authorship attestation, and amendment procedures.

The protocol is designed for implementation through the Secretary Suite system, but it may also be followed manually. Its central principle is:

> Declare the route, preserve the route, update the route, and accept responsibility for where the route leads.

1. Introduction

Authorship disputes often begin after the evidence needed to resolve them has disappeared.

A finished manuscript exists.

Its developmental history does not.

The publisher sees polished prose but cannot determine:

where the idea began;

whether the author developed an outline;

whether another person drafted the text;

whether AI produced substantial passages;

whether sources were verified;

whether revisions were directed by the named author;

or whether the author understands and accepts the final claims.

The finished artifact has become separated from its route.

That separation creates a provenance problem.

Provenance is the history of origin, development, transfer, and custody.

Chain of custody is the record preserving continuity across that history.

In authorship, the relevant chain is intellectual rather than merely physical.

It should answer:

> How did this idea become this book?

2. Authorship as a Route

A book does not appear in one step.

It travels through successive states:

\[

\text{intuition}

\rightarrow

\text{note}

\rightarrow

\text{question}

\rightarrow

\text{outline}

\rightarrow

\text{research}

\rightarrow

\text{draft}

\rightarrow

\text{criticism}

\rightarrow

\text{revision}

\rightarrow

\text{approval}

\rightarrow

\text{publication}.

\]

Each transition may involve different people, tools, and decisions.

The final language is therefore not the complete history of the work.

It is the endpoint of a route.

A valid provenance system must preserve enough of that route to distinguish:

origination from assistance;

suggestion from acceptance;

drafting from approval;

editing from replacement;

correction from concealment;

and publication responsibility from mechanical production.

3. Multiple Legitimate Routes

The protocol begins by rejecting the idea that all authentic books must follow one method.

Route A: Independent solo authorship

The author plans, researches, drafts, edits, and publishes without substantive generative or human writing assistance.

Route B: Conventional edited authorship

The author drafts the work and later works with developmental, line, copy, or proof editors.

Route C: Research-assisted authorship

Researchers gather evidence, conduct interviews, verify claims, or prepare background material under the author’s direction.

Route D: Ghostwritten authorship

The subject or directing author supplies the experiences, knowledge, objectives, and approval while a ghostwriter constructs substantial language.

Route E: Collaborative authorship

Multiple identified authors jointly develop the concepts and text.

Route F: AI-assisted authorship

The author uses AI for defined functions such as:

dialogue;

brainstorming;

organization;

critique;

comparison;

research assistance;

summarization;

drafting;

rewriting;

formatting;

citation assistance;

or voice refinement.

Route G: Mixed authorship

Several human and computational contributors perform separate roles.

The protocol does not rank these routes.

It records them.

4. The Authorship Process Declaration

Every project should begin with a dated Authorship Process Declaration.

This declaration identifies the anticipated route before the final work exists.

A general form is:

> Authorship Process Declaration — Version 1

This work originates with [name]. The originating author will direct the project and retain final approval and responsibility. The anticipated process includes [solo drafting / AI assistance / ghostwriting / editing / research assistance / collaboration]. The intended roles of each contributor or system are described below. This declaration will be updated if the process changes.

The declaration should contain:

project title or working title;

originating author;

date and time;

purpose;

expected audience;

anticipated contributors;

anticipated tools;

intended AI roles;

intended human-assistance roles;

disclosure level;

owner of final approval;

and process version.

The initial declaration is not a permanent restriction.

It is a starting record.

5. Amendment Rather Than Concealment

Creative methods change.

An author may begin alone and later hire an editor.

A project may begin without AI and later use AI for structural comparison.

A ghostwriter may add a researcher.

A publisher may require a translator.

The ethical requirement is not that the original plan never change.

The requirement is that the record change with it.

An amendment may state:

> Authorship Process Declaration — Version 2

On August 14, 2026, AI assistance was added for chapter-outline comparison and source organization. No AI-generated prose will be inserted without author review and revision.

A later amendment might state:

> Version 3

A developmental editor was added to evaluate continuity and argument order.

Changing routes is not deception.

Undocumented change creates the risk of deception.

6. The Originating Note

The chain should preserve the earliest identifiable expression of the work.

The originating note may be:

a typed paragraph;

a handwritten page;

a voice recording;

an email;

a conversation excerpt;

an image;

a diagram;

a question;

or a brief declaration of intent.

The note need not contain the whole theory or story.

Its purpose is to establish:

what first existed;

who expressed it;

when it was expressed;

and how the later work relates to it.

A useful originating note records:

1. The initial idea

2. The reason it matters

3. The intended direction

4. Known uncertainties

5. The anticipated next step

7. The Outline as Evidence

An outline is not merely a writing aid.

It is evidence of conceptual architecture.

A preserved outline shows that the author:

selected the major subjects;

arranged their relationships;

anticipated the sequence;

distinguished central claims from supporting material;

and developed the work before its final prose appeared.

The protocol should preserve:

the first outline;

every major revision;

added sections;

deleted sections;

reordered sections;

and reasons for substantial changes.

The outline may develop through stages:

\[

O_1

\rightarrow

O_2

\rightarrow

O_3

\rightarrow

O_f.

\]

Where:

\(O_1\) is the originating structure;

intermediate versions show development;

\(O_f\) is the final structural plan.

This chain is particularly valuable when AI or a ghostwriter assists with prose because it preserves evidence of the directing architecture.

8. The Source Record

A source record identifies the materials informing the work.

It should include:

title;

author;

publication;

date;

location or identifier;

date accessed;

relevant section;

claim supported;

verification status;

and whether AI located or summarized the source.

A source log should distinguish:

source text;

author interpretation;

AI summary;

independent verification;

and final use.

This prevents a common failure in AI-assisted research: a generated summary becoming detached from the source it supposedly represents.

The record should state when:

the source was read directly;

only an abstract was consulted;

the source was inaccessible;

the claim remains provisional;

or a citation was suggested but not verified.

9. The Contribution Log

A contribution log records who or what proposed each meaningful addition.

The log should distinguish:

9.1 Origination

Who first introduced the idea?

9.2 Expansion

Who developed it further?

9.3 Linguistic formulation

Who proposed the wording?

9.4 Selection

Who chose among alternatives?

9.5 Revision

Who changed the selected material?

9.6 Verification

Who confirmed its accuracy?

9.7 Approval

Who authorized inclusion in the final work?

A simple record might state:

Entry Contributor Proposed contribution Author decision Final status

014 AI assistant Three alternative chapter titles Modified by author Accepted

015 Author Commitment-state concept Expanded through dialogue Accepted

016 Editor Remove redundant example Approved Accepted

017 AI assistant Unsupported factual claim Rejected by author Excluded

The purpose is not to record every comma.

It is to preserve meaningful intellectual custody.

10. Suggestion Is Not Adoption

A system may generate twenty ideas.

The author may reject nineteen.

The one selected idea may then be rewritten beyond recognition.

A provenance system must not treat every suggestion as equivalent authorship.

The record should distinguish:

\[

\text{proposal}

\neq

\text{adoption}.

\]

And:

\[

\text{adoption}

\neq

\text{unchanged use}.

\]

Possible statuses include:

proposed;

rejected;

accepted;

accepted with modification;

merged;

superseded;

retained as background;

or removed after review.

This preserves the author’s judgment as an observable part of the process.

11. Draft and Version History

Every meaningful draft should receive:

project identifier;

version number;

date;

time;

contributor;

summary of changes;

source relationship;

and approval state.

A possible version structure is:

\[

1.0

\]

Initial full draft

\[

1.1

\]

Minor revisions

\[

2.0

\]

Major structural change

\[

2.1

\]

Post-review correction

\[

3.0

\]

Publication candidate

\[

3.1

\]

Published correction

Prior versions should not be overwritten when preserving them is reasonably possible.

A record can be maintained manually, through cloud history, through repository systems, or through dedicated provenance software.

12. Persistent Identity and Verification

Each project should receive a persistent identifier.

Each version may receive:

a unique version ID;

a timestamp;

a cryptographic hash;

a parent-version reference;

contributor identities;

and a status signature.

A hash does not prove that a claim is true.

It can help prove that a specific file existed in a specific form and has not been silently altered after the record was created.

The chain should therefore record:

\[

H(V_n)

\]

for each preserved version \(V_n\), where \(H\) is a cryptographic hash function.

The next version references the prior one:

\[

V_{n+1}

\rightarrow

H(V_n).

\]

This produces a verifiable sequence.

13. The Three Disclosure Layers

Not every reader needs access to every private draft.

The protocol therefore distinguishes three layers.

13.1 Public authorship statement

A concise statement included in the book, article, or publication page.

Example:

> This work originated with and was directed by John Swygert. AI was used for structured discussion, organization, research assistance, criticism, drafting support, and editorial refinement. The author selected, revised, approved, and accepts responsibility for the final work.

13.2 Verifiable project record

A more detailed record available to publishers, reviewers, institutions, collaborators, or interested readers.

It may include:

declaration versions;

outline history;

contributor roles;

version milestones;

source log summary;

and final attestation.

13.3 Private evidentiary archive

The complete confidential record.

It may include:

conversations;

voice files;

unpublished drafts;

notes;

correspondence;

contracts;

raw source materials;

rejected passages;

metadata;

and hashes.

The three-layer system protects both transparency and privacy.

14. The Final Authorship Attestation

Before publication, the named author should complete a final attestation.

A model statement is:

> Final Authorship Attestation

I have reviewed and approved this work in its final form. I understand its principal claims, arguments, and contents. I accept responsibility for its publication. To the best of my knowledge, the accompanying authorship and contribution record accurately describes the process through which the work was created.

The attestation may also identify:

known limitations;

unresolved factual questions;

rights status;

AI-generated components;

external contributor permissions;

and publisher disclosures.

The final attestation converts publication into an explicit act of responsibility.

15. Correction After Publication

Provenance should continue after release.

Corrections may be necessary because of:

factual error;

citation error;

omitted credit;

newly discovered evidence;

updated law;

revised data;

or process-disclosure correction.

A corrected edition should preserve:

original publication date;

correction date;

description of change;

responsible party;

new version identifier;

and relationship to the prior edition.

The chain should not pretend that the earlier version never existed.

Correction strengthens integrity when the correction itself is preserved.

16. Privacy and Proportionality

A provenance protocol should not become compulsory surveillance of thought.

Authors may need to protect:

confidential sources;

private medical information;

family communications;

trade secrets;

unpublished inventions;

security-sensitive material;

and intimate drafts.

The principle should be proportionality.

The record must preserve enough continuity to authenticate the route without forcing every private detail into public view.

Possible protections include:

sealed records;

redacted logs;

delayed release;

third-party escrow;

access permissions;

pseudonymous contributor IDs;

and selective verification.

Provenance should protect creation, not punish privacy.

17. Protocol Levels

The system may be implemented at three levels.

17.1 Minimum protocol

initial declaration;

originating note;

outline;

final manuscript;

public disclosure;

final attestation.

17.2 Standard protocol

declaration history;

outline versions;

source log;

major drafts;

contribution log;

public and verifiable records;

final attestation.

17.3 Enhanced protocol

complete version history;

cryptographic hashes;

timestamped conversation records;

contracts;

private archive;

third-party verification;

and persistent publication ledger.

The appropriate level depends upon:

commercial value;

institutional requirements;

controversy risk;

research significance;

number of contributors;

and the author’s preferences.

18. Application to Ghostwriting

A ghostwritten project should identify:

whose experiences and ideas form the work;

who conducted interviews;

who drafted prose;

who determined structure;

who revised;

who approved;

and what attribution agreement governs publication.

The public statement may remain brief if contract and convention permit.

The private record should be clear.

The honor of the arrangement does not require public performance of creative purity.

It requires that the involved parties tell the truth to one another and meet the disclosure obligations owed to publishers, readers, institutions, or law.

19. Application to AI-Assisted Writing

The Authors Guild has recommended that authors distinguish among kinds of AI use, fact-check outputs, comply with contractual obligations, and disclose substantial text generation. It also advises written agreements requiring editors, ghostwriters, and other contributors to reveal their own AI use. 

The Provenance Protocol extends that principle.

AI use should be described by function rather than by a binary yes-or-no label.

Possible functions include:

ideation;

interrogation;

outlining;

summarization;

source discovery;

comparative analysis;

draft generation;

rewriting;

style development;

translation;

formatting;

code generation;

image creation;

and quality control.

The record should also identify:

whether output entered the final work;

whether it was revised;

whether claims were verified;

whether the author understood it;

and who approved it.

20. Secretary Suite Implementation

Secretary Suite can implement the protocol as a persistent authorship layer.

Each project would contain:

Project identity

title;

author;

project ID;

purpose;

date opened;

status.

Authorship declaration

current version;

prior versions;

declared route;

contributors;

tools.

Origin record

first note;

recording;

transcript;

timestamp;

source identity.

Structural record

outline versions;

chapter map;

decision notes.

Source record

citations;

verification state;

claim mapping.

Contribution record

contributor;

role;

proposal;

author decision;

final status.

Version record

file;

parent version;

timestamp;

hash;

change summary.

Approval record

reviewer;

approval level;

comments;

final attestation.

Publication record

edition;

ISBN or DOI;

platform;

publication date;

correction history.

The system would preserve provenance when text is copied, exported, revised, or transferred between applications.

21. Provenance as Boundary

In TSTOEAO terms, provenance is a boundary that preserves identity through transformation.

The work changes:

\[

V_1

\rightarrow

V_2

\rightarrow

V_3.

\]

Without provenance, the later version may become detached from its prior states.

With provenance:

\[

V_3

\supset

\operatorname{history}(V_1,V_2).

\]

The record does not prevent change.

It ensures that change remains relationally accounted for.

This is dynamic equilibrium applied to authorship.

The project remains itself while evolving because the chain preserves continuity.

22. What the Protocol Does Not Prove

The protocol cannot prove that:

every idea was original;

every source was accurate;

every contributor was truthful;

the author fully understood everything;

the book is good;

or the claims are correct.

It can establish stronger evidence that:

the work developed through identifiable stages;

declared contributors participated in stated ways;

versions existed in a documented sequence;

the author made identifiable decisions;

and publication responsibility was accepted.

Provenance is not omniscience.

It is disciplined continuity.

23. The Ethical Principle

The protocol rests upon a simple distinction:

> Assistance is not deception. Concealment where truth is owed is deception.

An author should not be punished merely for using tools.

A reader should not be intentionally misled about material authorship where the representation matters.

A publisher should not have to rely exclusively upon intuition or unreliable AI detectors.

A genuine author should not lose a career because developmental evidence was never preserved.

The route should be available for examination at the level appropriate to the relationship.

Conclusion

Modern creation follows many routes.

Some authors work alone.

Some dictate.

Some collaborate.

Some hire ghostwriters.

Some use editors, researchers, translators, or artificial intelligence.

The future of authorship should not depend upon pretending that only one route is legitimate.

It should depend upon preserving truth across whichever route is chosen.

The Provenance Protocol begins before drafting with an Authorship Process Declaration.

It preserves:

\[

\text{origin}

\rightarrow

\text{outline}

\rightarrow

\text{research}

\rightarrow

\text{contribution}

\rightarrow

\text{draft}

\rightarrow

\text{revision}

\rightarrow

\text{approval}

\rightarrow

\text{publication}.

\]

It permits the process to change while requiring the declaration to change with it.

It distinguishes public transparency from private evidence.

It protects authors, collaborators, publishers, and readers.

Its governing principle is:

\[

\boxed{

\text{Declare the route. Preserve the route. Update the route.}

}

\]

\[

\boxed{

\text{Then accept responsibility for where the route leads.}

}

\]

References

Authors Guild. (2026). AI Best Practices for Authors. The Authors Guild. 

Loffhagen, E. (2026, July 31). $2m Crime Novel Deal Collapses Amid Questions Over AI Use. The Guardian. 

United States Copyright Office. (2025). Copyright and Artificial Intelligence, Part 2: Copyrightability. 

Swygert, J. (2026). The Tool Is Not the Lie: AI, Ghostwriters, Authorship, and the Ethics of Honest Creation. Ivory Tower Publishing.

The Tool Is Not the Lie: AI, Ghostwriters, Authorship, and the Ethics of Honest Creation

The Tool Is Not the Lie: AI, Ghostwriters, Authorship, and the Ethics of Honest Creation

DOI: Pending assignment

John Swygert

August 1, 2026

Abstract

Debates over artificial intelligence and authorship frequently begin with the wrong question. They ask whether a tool participated in producing the final language rather than asking who originated the work, directed its development, exercised judgment, preserved a coherent voice, approved the result, and accepted responsibility for publication.

Human creativity has never been tool-free. Writers have long relied upon editors, researchers, translators, stenographers, speechwriters, ghostwriters, dictionaries, recording devices, word processors, search engines, and publishing professionals. Artificial intelligence expands the speed and range of such assistance, but assistance alone does not erase human authorship.

This paper distinguishes authorship from mechanical production. It argues that authorship is a relationship of origination, direction, judgment, voice, understanding, approval, and responsibility. A tool supplies leverage. It does not automatically supply purpose or ownership.

The paper rejects the false standard of creative purity and proposes an integrity standard instead. Under that standard, multiple routes to authorship may be legitimate: solo writing, editorial collaboration, ghostwriting, research assistance, AI-assisted development, and mixed human-machine workflows. The ethical boundary is not whether assistance occurred. It is whether the route was truthfully represented, whether meaningful contributions were properly accounted for, and whether the named author retained intellectual custody of the work.

A recent publishing controversy involving Jerry Falade’s novel Call Me, I’ll Hide the Body illustrates the danger created when provenance cannot be authenticated. The failure was not merely a question of whether AI had been used. It was a failure of trust and evidentiary continuity concerning how the manuscript developed. This paper argues that future authorship should be supported by documented provenance: a traceable chain from originating idea through outline, drafts, assistance, revisions, approval, and publication.

The central proposition is:

> The author supplies purpose, direction, judgment, voice, approval, and responsibility. The tool supplies leverage. Integrity supplies legitimacy.

1. Introduction

A mechanic repairs an engine using diagnostic equipment, lifts, manuals, wrenches, measuring instruments, replacement parts, and computer systems.

Nobody asks:

> Did you really repair the car yourself, or did you cheat by using tools?

The question would sound absurd because the mechanic’s authorship of the repair does not depend upon touching every component with bare fingers.

The mechanic:

identifies the problem;

determines the desired outcome;

chooses the tools;

interprets the evidence;

rejects incorrect possibilities;

performs or directs the intervention;

checks the result;

and accepts responsibility if the repair fails.

The tools provide capability.

They do not independently establish purpose, judgment, or responsibility.

Writing should be analyzed through the same distinction.

A person may originate an idea, develop its conceptual architecture, dictate its substance, test it through dialogue, reject weak formulations, demand revisions, approve the final language, and place the resulting work into the public record under their own name.

The fact that an editor, ghostwriter, researcher, or artificial intelligence helped transform that thought into polished language does not automatically transfer authorship away from the directing mind.

The central ethical question is not:

> Was a tool used?

It is:

> Who governed the work, and was the route of creation represented honestly?

2. The Myth of Creative Purity

The myth of creative purity imagines the authentic author as an isolated person producing every idea, sentence, correction, citation, structural decision, and final artifact without outside assistance.

That person has rarely existed.

Language itself is inherited.

Every writer uses words developed by prior generations.

Every scholar uses concepts, methods, and notation built by other minds.

Every book depends upon some combination of:

education;

conversation;

cultural inheritance;

prior reading;

reference materials;

technologies of inscription;

manufacturing systems;

distribution systems;

and readers who preserve and reinterpret the work.

Even the supposedly solitary author normally relies upon:

spelling conventions;

dictionaries;

typesetting;

software;

editorial standards;

publishing formats;

and material tools.

Creative purity is therefore not the historical condition of authorship.

It is a romantic fiction.

A work can be deeply personal without being mechanically solitary.

A work can be authentically directed without being produced by one pair of hands.

A work can contain the voice of one author while having passed through many forms of assistance.

3. Authorship Is Not Typing

The act of typing is one route through which authorship may be expressed.

It is not authorship itself.

Authorship includes several distinct functions.

3.1 Origination

The author identifies the central idea, experience, problem, question, story, argument, or purpose.

3.2 Direction

The author determines where the work should go, what belongs, what does not belong, and what the finished work is intended to accomplish.

3.3 Judgment

The author distinguishes strong material from weak material, accurate claims from exaggeration, and useful expression from distortion.

3.4 Voice

The author supplies recurring patterns of perspective, emphasis, humor, restraint, anger, curiosity, morality, metaphor, and conceptual relation.

3.5 Understanding

The author understands the work well enough to defend, revise, explain, and correct it.

3.6 Approval

The author decides when the work accurately expresses the intended meaning.

3.7 Responsibility

The author accepts accountability for the final publication.

These functions are more important than the physical production of every sentence.

A person who types every word may still plagiarize another person’s ideas.

A person who dictates a book, works with an editor, or uses AI may still exercise complete intellectual direction.

Fingerprints on the keyboard are therefore an inadequate measure of authorship.

4. Ghostwriting Reveals the Double Standard

Ghostwriting has long existed as a recognized professional practice.

The Authors Guild openly maintains professional listings for ghostwriters, editors, writing coaches, and related specialists, demonstrating that such assistance is not an underground anomaly but an established part of the publishing ecosystem. 

A public figure, military officer, executive, physician, survivor, or political leader may possess:

the experience;

the memories;

the expertise;

the intended lessons;

the moral position;

and the reason for writing,

while lacking the time, training, health, or desire required to structure an entire commercial manuscript.

A ghostwriter may:

conduct interviews;

organize chronology;

identify narrative arcs;

construct prose;

clarify repetition;

reconcile conflicting recollections;

and create readable continuity.

The resulting work may still fundamentally belong to the person whose life, knowledge, purpose, and decisions govern it.

The ethical quality of the arrangement depends upon truthfulness, consent, contract, responsibility, and expectations—not upon the absence of assistance.

The double standard becomes obvious when society treats concealed or quietly accepted human ghostwriting as respectable while treating openly acknowledged AI-assisted articulation as inherently fraudulent.

The question should not be whether the assistant was human or computational.

The question should be:

> What role did the assistant perform, and how was that role represented?

5. Cognitive and Linguistic Authorship

A recent Psychology Today essay distinguishes between cognitive authorship and linguistic authorship.

Cognitive authorship concerns the judgment, reasoning, and selection that determine which idea matters and how it should be shaped.

Linguistic authorship concerns the wording and structure through which that idea is communicated. The essay argues that modern AI can produce fluent linguistic output even when the reader cannot determine whether equivalent human thought preceded the sentences. 

That distinction is useful, but the two forms need not belong to the same agent in equal proportions for legitimate authorship to exist.

An author may provide:

the cognitive foundation;

the lived experience;

the framework;

the objectives;

the conceptual corrections;

and the final authority,

while another person or system helps produce or refine the linguistic form.

This already occurs through:

dictation;

developmental editing;

translation;

speechwriting;

ghostwriting;

and collaborative research.

Artificial intelligence makes the separation more visible, but it did not invent it.

The proper question becomes:

> Where did the governing cognition reside?

When a user brings an AI a shallow prompt and accepts a complete generated work without understanding or revising it, the user’s claim to full authorship is weak.

When a user develops the ideas through prolonged dialogue, repeatedly corrects the system, establishes a body of concepts, rejects inaccurate language, shapes the argument, and approves the final text, the human role is substantially different.

The AI may contribute language and analysis.

The human may remain the directing author.

6. Voice Is More Than Sentence Style

An author’s voice cannot be reduced to a few stylistic features.

Voice includes:

what the author notices;

what the author refuses to overlook;

which distinctions matter;

how uncertainty is handled;

when humor interrupts seriousness;

how moral responsibility is assigned;

what metaphors recur;

which claims are considered too strong;

and what kind of explanation feels complete.

A voice may be developed through years of writing.

It may also be developed through extended conversation.

When an author and an AI discuss physics, health, politics, music, memory, technology, design, publishing, history, family, ethics, and daily life, the system receives more than isolated writing samples.

It receives a long record of:

correction;

preference;

conceptual insistence;

rejection;

emotional emphasis;

vocabulary;

humor;

and reasoning style.

The resulting assistance can increasingly reflect the author’s developed voice because that voice has been repeatedly demonstrated and defended.

An outsider reading the finished book may not see that process.

That does not mean the process did not occur.

It means that authorship needs provenance capable of making developmental continuity visible when necessary.

7. Efficiency Is Not Fraud

A persistent objection to AI-assisted creation is that the tool allows work to be completed too quickly.

That objection confuses effort with value.

A difficult task may take years because the available tools are primitive.

The same intellectual task may later take months, weeks, or days because better tools remove mechanical obstacles.

The reduction of unnecessary labor does not prove that thought disappeared.

A spreadsheet can complete calculations that once required teams of clerks.

Computer-aided design can accelerate engineering.

Digital audio systems can replace physical cutting and splicing.

Searchable databases can replace weeks of manual catalog work.

None of these improvements automatically destroys the intellectual ownership of the person directing the work.

Time consumed is not proof of authorship.

Suffering is not proof of originality.

Inefficiency is not integrity.

The proper standard is whether the accelerated process still contains:

thought;

understanding;

judgment;

verification;

revision;

responsibility;

and honest representation.

Why should an author spend three years performing clerical organization that can be completed in three months if the author can use the saved time to think more deeply, produce additional work, test more alternatives, or preserve ideas that might otherwise be lost?

Efficiency becomes dangerous only when it removes judgment rather than drudgery.

8. The Legal Distinction Between Assistance and Machine Determination

The United States Copyright Office has stated that using AI to assist in creation does not by itself prevent copyright protection.

Protection may exist where sufficient human creativity determines the expressive elements, including situations in which a human creatively arranges, modifies, or integrates AI-related material. Merely supplying prompts, however, does not automatically establish authorship over machine-determined expression. 

This is a legal distinction rather than a complete moral theory, but it reinforces an important principle:

> AI involvement and human authorship are not mutually exclusive.

The legal question turns partly upon which expressive elements were determined by the human.

The ethical question should extend further and examine:

who originated the work;

who controlled its architecture;

who selected among alternatives;

who made substantive revisions;

who understood the result;

and who assumed responsibility.

The law cannot fully measure voice, direction, intellectual custody, or moral responsibility.

Provenance can make those relationships more visible.

9. The Publishing Controversy as a Provenance Failure

In July 2026, the agency representing Jerry Falade’s debut crime novel Call Me, I’ll Hide the Body withdrew the manuscript after questions arose concerning possible AI involvement.

The book had reportedly attracted a fourteen-way auction and an offer exceeding $2 million. The agency said it could no longer authenticate how the manuscript developed from origin to completion. Falade denied using AI and argued that bias influenced the response. The agent continued to describe the manuscript as exceptional while explaining that trust in the account of its development had broken down. 

The public record does not establish with certainty how the manuscript was produced.

That uncertainty is precisely the point.

The crisis arose because the chain of authorship could not be confidently reconstructed.

The issue was not simply:

> Was AI used?

The larger issue was:

> Can the origin, development, assistance, and final responsibility for the manuscript be authenticated?

The book’s literary quality did not suddenly disappear.

What disappeared was institutional confidence in the production history.

A publishing house considering a multimillion-dollar acquisition must evaluate:

contractual warranties;

copyright risk;

reputational exposure;

public trust;

future adaptation rights;

and the author’s representations.

A provenance vacuum transforms uncertainty into financial danger.

The proper lesson is not merely “never use AI.”

It is:

> Do not allow the history of the work to become unknowable.

10. The Absence of Provenance Changes Probability

The absence of a complete archive does not prove fraud.

Some writers draft in a single file.

Some destroy notes.

Some lose devices.

Some compose orally.

Some rewrite entire sections without preserving prior versions.

Some processes are naturally less documented than others.

Nevertheless, when a highly valuable manuscript is involved, the absence of:

early fragments;

dated drafts;

outlines;

cloud history;

research notes;

correspondence;

reader feedback;

revision records;

or consistent explanation

may legitimately increase concern.

Provenance does not function only as a weapon against authors.

It protects honest authors from accusation.

A writer falsely accused of machine generation should be able to produce evidence of development.

A writer who openly used AI should be able to show where the ideas began, how the system was directed, what was accepted or rejected, and how the final work was governed.

The same record protects both.

11. Self-Publishing and Self-Authenticity

Self-publishing does not prove that a work was written without assistance.

It does preserve a meaningful form of self-authenticity.

The author:

chooses the title;

creates or approves the manuscript;

controls the publication account;

determines the editions;

authorizes revisions;

sets the publication date;

accepts public association with the work;

and assumes direct responsibility for placing it into circulation.

There is no publishing institution serving as the final gatekeeper.

The act of self-publication is therefore a direct declaration:

> I am placing this work into the public record under my own authority.

That is not complete evidentiary proof of every stage of production.

It is nevertheless a strong act of ownership and responsibility.

Traditional publishing can provide editing, distribution, prestige, marketing, and institutional support.

Self-publishing preserves sovereignty.

For an author whose central goal is to complete, preserve, and release the work, sovereignty may matter more than permission.

12. The False Test and the Proper Test

The false authorship test asks:

Did the author personally type every word?

Did the author use no generative tool?

Did the author work entirely alone?

Did the process take sufficiently long?

Did the author endure enough difficulty?

Is every sentence traceable to unaided spontaneous composition?

The proper test asks:

Who originated the central work?

Who directed its development?

Who contributed language, research, structure, or revision?

Was the assistance represented honestly?

Did the named author understand the work?

Did the author exercise meaningful selection and rejection?

Did the author preserve a coherent voice?

Did the author approve the final artifact?

Who accepts responsibility?

Can the work’s developmental chain be reasonably authenticated?

This is the distinction between purity and integrity.

Purity rejects assistance.

Integrity accounts for assistance.

Purity romanticizes isolation.

Integrity preserves truth.

Purity asks whether tools touched the work.

Integrity asks whether deception entered the work.

13. Multiple Legitimate Routes to Authorship

There is no single legitimate route.

A work may be:

13.1 Solo-authored

One person originates, drafts, edits, and publishes the work.

13.2 Editor-assisted

The author writes the manuscript while an editor improves structure, clarity, accuracy, or style.

13.3 Research-assisted

The author directs the work while researchers gather evidence, verify claims, or organize sources.

13.4 Ghostwritten

The named author supplies experience, knowledge, direction, and approval while a ghostwriter constructs substantial prose.

13.5 Collaboratively authored

Multiple humans jointly create the ideas and language.

13.6 AI-assisted

The human uses AI for brainstorming, criticism, research assistance, organization, drafting, rewriting, comparison, formatting, or other defined roles.

13.7 Mixed-route authorship

Human authors, editors, researchers, ghostwriters, and AI systems contribute through separately identified functions.

None of these routes is inherently honorable.

None is inherently dishonorable.

Their legitimacy depends upon consent, control, truthfulness, attribution, and responsibility.

> The route does not determine integrity. Truthfulness about the route does.

14. A Practical Integrity Standard

A modern author should be able to state:

1. Where the work began

2. Who originated its central ideas

3. Which people or systems contributed

4. What roles they performed

5. How the work changed

6. Who selected and approved the final form

7. Who accepts responsibility

8. What evidence preserves the development

This does not require publishing every private conversation.

It requires preserving enough of the route that meaningful claims can be authenticated.

A brief public statement may be sufficient for ordinary readers.

A detailed private archive may be retained for disputes, publishers, courts, institutions, or future historians.

The purpose is not surveillance.

The purpose is continuity.

15. Secretary Suite and the Future of Authorship

Secretary Suite proposes that AI-assisted text carry persistent authorship provenance.

The system would preserve:

human author identity;

assisting agent identity;

date and time;

project purpose;

version;

source material;

contributor roles;

edit history;

approval history;

and verifiable chain of custody.

Copied text should not become detached from its origin.

Revised text should not erase its prior stages.

A final book should be traceable backward through:

\[

\text{publication}

\rightarrow

\text{final approval}

\rightarrow

\text{revisions}

\rightarrow

\text{drafts}

\rightarrow

\text{outline}

\rightarrow

\text{originating note}.

\]

This transforms authorship from an unverifiable purity claim into a documented history of intellectual custody.

The system would not declare one route morally superior.

It would allow the route to be seen.

Conclusion

The rise of artificial intelligence has not destroyed authorship.

It has exposed how poorly authorship was previously defined.

Society often treated typing, thinking, wording, originality, ownership, and responsibility as though they were one indivisible act.

They never were.

Ghostwriters, editors, researchers, collaborators, and speechwriters have long separated these functions.

AI makes the separation faster, more visible, and more accessible.

The ethical response should not be to demand tool-free creation.

It should be to require truthful creative custody.

A person remains meaningfully authorial when that person originates or governs the work, shapes its conceptual development, exercises judgment, preserves voice, understands the result, approves the final form, and accepts responsibility.

A person weakens the claim to authorship when the work is accepted without understanding, falsely represented, materially copied, or detached from any meaningful human direction.

The proper boundary is not:

\[

\text{tool}

\quad\text{versus}\quad

\text{no tool}.

\]

It is:

\[

\text{honestly directed creation}

\quad\text{versus}\quad

\text{misrepresented production}.

\]

The mechanic is not a fraud because a wrench touched the engine.

The writer is not a fraud because an intelligent tool helped shape expression.

The lie begins only when the route is concealed where truth is owed.

The governing principle is:

\[

\boxed{

\text{The author supplies purpose, direction, judgment, voice, approval, and responsibility.}

}

\]

\[

\boxed{

\text{The tool supplies leverage. Integrity supplies legitimacy.}

}

\]

References

Authors Guild. (2026). AI Best Practices for Authors. The Authors Guild. 

Authors Guild. (2026). Ghostwriters: Writers Marketplace. The Authors Guild. 

Loffhagen, E. (2026, July 31). $2m Crime Novel Deal Collapses Amid Questions Over AI Use. The Guardian. 

Nosta, J. (2026, July 30). AI and the Two Forms of Authorship. Psychology Today. 

Taylor, J. (2026). When AI Does the Writing, Who’s the Author? Psychology Today. 

United States Copyright Office. (2025). Copyright and Artificial Intelligence, Part 2: Copyrightability.