Ethics, Rights & Disclosure
Before we touch a single generator, we deal with the part nobody puts in the launch video: who owns what AI makes, whose consent you need, and what you owe your collaborators and your audience. Get this chapter wrong and no tool in the course can save you.
This is the chapter I'd teach first if I could only teach one. The tools will change under our feet — half the products in this course will have new names in two years — but the questions in this chapter don't change: who owns it, who consented to it, and who deserves to know. Every AI decision on a film runs through those three, and I want them installed in your head before you generate your first frame.
Start with ownership, because it surprises people. Under current U.S. copyright doctrine, purely AI-generated material generally can't be copyrighted — copyright protects human authorship, and a machine's output with no meaningful human creative contribution doesn't qualify. Think through what that means practically: if a shot in your film is straight generator output, you may not be able to stop anyone else from using something essentially identical. Your film as a whole — the writing, direction, editing, the human choices — remains protectable, but the generated asset itself may sit in a gray zone. The more you shape, composite, and rework generated material with your own authorship, the stronger your position. The doctrine is evolving, so verify the current state before you rely on any of this.
The rules you don't break
- Never clone a voice or likeness without written consent. Not a dead actor's, not your lead's, not your friend's "as a joke." Written, specific, on file. No exceptions.
- Don't assume you own generated output. Purely AI-generated material generally isn't copyrightable; know which assets in your film sit in that gray zone before a distributor asks.
- Check the union agreements before AI touches a performer. SAG-AFTRA has negotiated consent and compensation requirements around digital replicas and AI voice work — read the current agreement that applies to your production, because the terms keep evolving.
- Disclose when they ask — and honestly. Festivals, distributors, and grant bodies increasingly ask what AI touched. Answer straight. Getting caught in an omission costs more than the disclosure ever would.
- Tell your collaborators. If AI-generated material shares the frame with an actor's performance or a designer's work, they find out from you, before it happens — not from the credits.
Behind all of this sits the training-data controversy, and you should understand it even in brief: many generative models were trained on scraped creative work — scripts, images, films — without the creators' permission, and the lawsuits over whether that's infringement are still working through the courts. I won't pretend that's settled, and I won't pretend it's simple. What I'll say is this: you're a creator using tools that may have been built partly on other creators' unconsented work. Sit with that honestly. For some filmmakers it's a reason to prefer tools trained on licensed data; for others it's a reason to keep AI away from anything expressive and use it only for logistics. Both are defensible positions. Pretending the question doesn't exist is not.
Disclosure is a craft position, not a confession
Here's the reframe that took me too long: disclosure isn't an admission of guilt, it's a statement of process — the same way "shot on location" or "practical effects" is. When a festival form asks whether AI was used, the filmmakers who answer plainly ("AI-generated concept frames in the pitch deck; all final footage photographed") read as professionals in control of their pipeline. The ones who fudge it read as people who think they did something wrong. And functionally, they did — because the cover-up converts a normal tool choice into a trust problem with the exact people deciding whether to program, fund, or distribute your film.
The same logic applies on set. Your actors are watching this technology with real fear, some of it earned. When you're upfront — this is what AI touches on this production, this is what it will never touch, here's the consent paperwork — you're not slowing the production down. You're buying the kind of trust that gets you the performance, and gets your next film crewed by people who vouch for you.
An actor once asked me, halfway through a take day, whether the temp voiceover in the edit would be "some AI version of me." Nobody had told her we were using a scratch synthetic voice — not hers — for timing. Twenty minutes of honest conversation and a written note in her agreement fixed it. But I watched the trust drain out of the room in the minute before that conversation started, and I don't ever want to watch it again. Tell people first.
One last thing, and it matters: I'm a filmmaker, not a lawyer, and none of this is legal advice. The doctrine, the union terms, and the disclosure norms are all moving. When real money, real distribution, or a real person's likeness is at stake, talk to an entertainment lawyer before you commit. It's the cheapest insurance on your budget.
You're already using AI when you use the Filmmaker Toolbox — script breakdowns, shot lists, budgets, and pitch decks, generated in minutes. It's the producer side of everything this course covers, built for filmmakers.
