Should You Disclose AI Use in Your Script?

Should you disclose that you used AI on a script? The question makes people jumpy, and the jumpiness produces two bad answers. The panic answer says never touch the tool, so there is nothing to disclose, which ignores that the tool is already on every desk. The lazy answer says never mention it, take the credit, keep quiet, which is how trust dies quietly. The honest answer is more useful than either, and it turns on a single distinction most people skip. Let me lay it out the way I would to a writer about to send a script into the world.

The distinction that decides everything

Before you ask whether to disclose, ask what the machine actually did. There are two very different uses hiding under one scary word. In the first, the machine assisted: it formatted your pages, caught a plot hole, broke a logline six ways so you could pick the live one. In the second, the machine manufactured the thing that is supposed to be yours: it wrote the emotional beat, generated the voice, produced the felt truth you then passed off as your own lived feeling. These are not two shades of the same act. They are different acts, and they carry different obligations. The first is using a tool, like a calculator or a spellchecker. The second is closer to a small fraud on the audience, the one I drew a hard line around in the larger argument.

For assistance, disclosure is a courtesy, not a confession

If the machine only assisted, you are not hiding a crime. Nobody expects a novelist to disclose that they used a spellchecker or a thesaurus. The relevant question becomes practical, not moral: does the person receiving the work want to know, and does the contract require it. Increasingly, contracts do. Productions and publishers are adding AI clauses, and the honest move is simple: read the contract, answer it truthfully. The Authors Guild's guidance leans toward transparency as the safe default, and I agree, because a workflow you would be embarrassed to describe is usually a workflow worth fixing. But assistance disclosed is not a stain. It is just the truth, told plainly, about a tool.

Disclosure is not a confession that you cheated. It is a promise that you did not.Wr. Sarkhedi

For the felt part, silence is the real problem

Now the harder case. If you let the machine manufacture the emotional truth and stayed quiet, disclosure is not really the issue. The problem happened earlier, when you passed off an assembled feeling as a lived one. Readers are strange, sensitive animals, built to sense whether the person across the fire actually feels what they claim. That is why the felt part of the craft cannot be faked for long. The dishonesty is not in failing to disclose. It is in there being something to disclose in the first place, a hollow center wearing your name. Fix that and the disclosure question mostly dissolves, because you will not have handed the machine the one job that was yours.

What the industry has already decided

You are not writing into a vacuum on this. When screenwriters organised in 2023, one of the guardrails they set was that AI cannot be credited as a writer and cannot be used to quietly strip a human of credit or pay. Read plainly, that is an industry saying: the human authorship must be real and visible, and the machine's role must not be laundered into a person's credit. The principle travels even if you never sign a guild card. Be able to say what the tool did. Do not let its work be mistaken for a person's, including your own. Those tracking the pressure on new writers, like the writers warning about the vanishing entry-level bridge, are pointing at the same underlying value: keep the human real and named.

A scene where the distinction actually gets tested

Let me invent a situation to show how this plays out under real pressure, not just in principle. Imagine a writer named Kunal finishing a pilot script on a tight deadline, a family drama about two brothers dividing their late father's small manufacturing business. He is three days from the table read and stuck on the scene where the elder brother realizes the younger one has been quietly siphoning funds for years to pay a debt he never disclosed. Kunal runs the scene through a tool twice: once to test whether the ledger discovery should happen on page, and once to check whether a piece of dialogue about the father's will sounds plausible for a small-town accountant character. Neither use touched the emotional center of the confrontation, which Kunal wrote from a specific memory of watching his own uncles divide a shop after his grandfather died. If a producer asked Kunal what he used AI for on this script, he could answer in one sentence, truthfully, without flinching, because nothing in his answer touches the part of the scene that actually carries the story's weight.

Now imagine a second version of Kunal, equally rushed, who instead asks the tool to write the confrontation itself, likes the shape of the dialogue, and ships it without ever interrogating why the elder brother reacts the way he does. If asked the same question by the same producer, this Kunal has a much harder sentence to say out loud, not because disclosure rules are strict, but because the thing he would be disclosing is that the emotional center of his own pilot was not his. The discomfort in that second scenario is the actual signal. It is not telling you to phrase your disclosure more carefully. It is telling you the draft is not finished being written by you.

Building the habit before it is tested

The easiest way to never dread this question is to build a habit long before a producer or a contract asks it. Keep a running, honest note to yourself, even an informal one, of what a tool touched on a given draft: structure check, research, line polish, nothing more. If the note ever needs to include "wrote the emotional core of the confrontation scene," treat that as a flag to go back and write that scene yourself before the draft goes anywhere. This is a smaller, private version of the same discipline I described for outlining in the piece on outlining a screenplay: know what you built and why, so that when someone asks, you are reporting a decision rather than reconstructing an excuse.

The practical rule I would give a younger writer

Here is the whole thing compressed into rules you can actually carry:

• Read every contract for AI clauses and answer them honestly. When asked, tell the truth. Always.
• If the machine only assisted, disclosure is easy and costs you nothing but a sentence.
• If the machine made the felt part, do not disclose it, remove it. Write that part yourself. That is the fix, not the confession.
• Never let the tool take a credit or a payment that belongs to a person. The same principle runs through the wider ethics of AI-assisted writing, and it does not bend for a tight deadline.

The question is not whether to admit you used it. It is whether you would be ashamed once you did.

Transparency is not the risk. Hiding is.

Writers fear that disclosing AI use will make their work look lesser. In the short run, maybe, in a nervous market. In the long run, the opposite is true. As machine-made pages flood everything, the writers who are open about their process and clear about what is genuinely theirs will be the ones trusted, and trust is about to be the scarcest currency in the business. The ones who hid it, and got caught with a hollow center where a feeling should have been, will not get a second read. So do not think of disclosure as a tax on using the tool. Think of it as the habit that keeps your name meaning something, which, in the years ahead, is the only job security a writer has left. This is also, in a quieter way, a question about karma in the oldest sense of the word, the idea that an action carries its consequence inside itself rather than in whether anyone else notices. Tell the truth about the tool. Keep the feeling yours. That is the entire answer, and if you are still deciding whether the tool belongs on your desk at all, that larger question is answered plainly in the honest test of what AI can and cannot write.