Sunny Greenhill – How to Write a Books with ChatGPT (страница 1)

Sunny Greenhill

How to Write a Books with ChatGPT

Chapter 1. The Legal Reality of Co-Authorship with AI

When a person first writes together with AI, they almost always experience a strange feeling… as if they have sat down at a table with someone very intelligent, someone who speaks quickly and confidently and, for some reason, is willing to produce text in batches. That feeling is deceptive. Not because AI is "bad," but because it pushes you toward the most dangerous thought: "Well, if it wrote it, then it is the author… or at least the co-author."

And here I want to stop you, kindly but firmly. Because a writer can afford mysticism and ambiguity in a plot… but in the legal part, ambiguity turns into risk. The risk of losing rights, the risk of running into a publishing ban, the risk of receiving a claim, and finally, the risk of one day finding yourself looking at your own book and being unable to explain with confidence: "why is this mine."

This chapter is not about "how to think correctly about AI." It is about how the world of documents and contracts will think about you. That world is old. It likes clarity. It likes wording. It likes provability. And it does not like the word "co-author" where the co-author is not a human being.

We are going to move calmly, step by step. First, the foundation: what authorship actually means in the legal sense and why AI does not fall into that category. Then it gets more difficult: the gray areas, where everything starts being interpreted differently in different countries and by different platforms. And then practice: how to build your process so that your authorship is not "by feeling," but by fact.

1.1. Who Is Considered the Author When AI Participates in the Process

Let us begin with this: legal authorship is not a badge of recognition and not a matter of ego. It is a dry, utilitarian construct that determines one thing: who has the right to control the text – to publish it, sell it, transfer it, prohibit its use, and bear responsibility for it.

In this sense, the law is like passport control. It is not interested in "how you feel about yourself." It cares about "who you are." And here the first fundamental thing appears, something you need to understand once and for all:

AI cannot be an author not because it "does not know how," but because it is not a legal subject.

What does "not a legal subject" mean? It means this: it does not have legal status the way a human being or a company does. It cannot go to court. It cannot sign a contract. It cannot own property rights. It cannot bear responsibility. It cannot "say": "this is mine and I forbid it" – and then be obligated to prove and defend that.

So legally, AI is a tool. A very complex one, sometimes resembling an intelligent conversation partner, but in law, those similarities do not count as arguments.

Now the next question, the one a beginning author always asks: "All right, AI is not the author. So the author is me? Even if the draft text was written by it?"

And here the important nuance begins. In law, the author is not the one who physically typed the greater number of letters. The author is the one who created the protected form of expression.

And form of expression is not only words. It is:

●

how the scene is built,

●

in what order the events unfold,

●

what rhythm the sentences have,

●

where you place pauses,

●

what intonation the character has,

●

how the composition of the chapter works,

●

what you threw out and what you kept.

Let me try to explain it in human terms.

Imagine a movie. The camera has shot footage. There is a lot of it. It is raw. There are good pieces, there are failures, there are alternate takes. The mere existence of video does not make it a film. A film appears in the editing: when someone selected the shots, arranged them, built the tempo, cut out the excess, added semantic connections, and made everything work as a single work.

The story with text and AI is very similar.

AI can produce the "shot footage" – paragraphs, variants, a draft of a scene. But the work appears where a human being takes on direction and editing. And that word – control – is the main axis of legal authorship in the human-AI relationship.

If you control the process in such a way that you:

●

set the direction,

●

make decisions,

●

go through alternatives,

●

rework,

●

bring it to a coherent authorial sound,

then you develop what legal logic calls a "human creative contribution." And that is what makes you the author, and AI the tool.

But if the process looks different… if you act like an operator rather than a director, then you enter the zone where disputes begin.

This is what it usually looks like.

A beginning author makes a request: "write a chapter," gets 15,000 characters, and thinks: "well, it is fine, let it stay that way." They change a couple of words so that it is "not quite like AI," put their name on it, and publish it. And if a legal question arises in that situation, their position will be weaker because it is hard to show: where exactly is the human creativity here?

Do you see the trap? AI makes the text too "ready," and the beginner stops working like an author. They start working like a consumer.

I will say something unpleasant but honest: in the legal world, "I spent time" does not equal "I created a work."

You can spend three hours running prompts… and still make not a single creative decision. And you can spend one hour aggressively re-editing a scene, rewriting dialogue, changing perspective, building a chapter – and that will be more significant than all the "automatic" labor.

Now an important point: in different countries and systems, the attitude toward this question may differ. In some places, the requirement for human creative contribution is stricter; in others, softer; in some places, they focus more on the fact that the person "organized" the process. But the general principle is the same everywhere: if the text does not show visible human authorship, risk arises.

The risk is not that "your book will be taken away." The risk is something else:

●

a publisher may not want to get involved,

●

a platform may request confirmation,

●

in a conflict, you will have fewer levers,

●

questions may arise when selling rights,

●

and most importantly – your position becomes hazy.

And haze, remember, is always more dangerous in contracts than bad style.

Now – what should you do to get out of this trap?