Using AI Images Responsibly: Copyright, Likeness, and Disclosure

Start with an example

In early 2024, an AI-generated video circulated online showing a well-known musician “endorsing” a cookware giveaway. It looked and sounded real. It wasn’t. The giveaway was a scam, and fans who clicked lost money. The musician had never agreed to anything.

Around the same time, a small business owner on social media proudly posted an AI image of a famous cartoon character holding her product. Within a week she had a takedown notice from the company that owns the character.

Neither of these people was a criminal mastermind. One case was fraud; the other was simple carelessness. But both show the same truth: AI makes it easy to create images of things you don’t have the right to use. This lesson teaches you where the lines are.

Note: This lesson explains general principles. It isn’t legal advice. Laws are changing quickly, and for specific situations, talk with a qualified attorney.

Part 1: Who owns an AI image?

What the U.S. Copyright Office says

Copyright protects original creative work, and in the U.S., it requires human authorship. The U.S. Copyright Office has said that material generated entirely by AI, such as an image produced from a text prompt, generally can’t be copyrighted on its own. Courts have agreed so far.

But that doesn’t mean nothing is protected. Your human contributions can be, such as:

• Substantial edits you make by hand to an AI image
• The creative arrangement of AI images with your own text, photos, and design in a flyer or book
• Your own original photos or drawings that you then edit with AI

What this means for you: If your logo or mascot is purely AI-generated, others may be able to copy it legally. For important brand assets, have a designer refine or redraw the AI concept so there’s real human authorship.

Check the tool’s terms

Every AI tool has terms of service that explain what you’re allowed to do with the images it makes, including commercial use. Most allow business use, but some free plans have limits. Read the terms before you print or sell anything.

Part 2: Don’t copy others’ work

AI models learned from enormous numbers of images, and they can reproduce recognizable characters, logos, and artistic styles. Just because a tool can make something doesn’t mean you’re allowed to use it.

Avoid generating:

• Famous characters: cartoon mascots, superheroes, video game characters
• Brand logos and trademarks: sports teams, sneaker brands, restaurant chains
• Specific copyrighted artworks: recreating a famous painting or album cover
• A living artist’s signature style by name: “in the style of [artist]” is legally uncertain and disrespects the artist’s work

Instead, describe qualities: “bold outlines, flat bright colors, and playful exaggerated proportions.” You get the feel you want while creating something original, and you build your own creative voice.

Part 3: People’s faces and voices

A person’s likeness (their face, voice, and identity) belongs to them. Many states protect it through right of publicity laws, which stop businesses from using someone’s identity to sell things without permission.

Illinois has one of the strongest of these laws, and in 2024 it was updated to address AI-created digital replicas of people’s likeness and voice. Other states, including Tennessee and California, have passed similar laws. A federal law, the TAKE IT DOWN Act, now requires platforms to remove non-consensual intimate images, including AI-generated ones.

The rules for you:

Situation	What to do
Creating an image or video of a real, identifiable person	Get written permission that says how it will be used
Editing photos of customers or employees	Get permission; respect “no”
Images involving children	Get a parent or guardian’s written consent, and be extra cautious
Creating a “spokesperson”	Use an original character, like R.I.C.K. Bot, instead of a real person’s likeness
Voice cloning	Only your own voice, or someone’s with their explicit, written consent

Never create AI images that sexualize, humiliate, or impersonate real people. Beyond being harmful, it can be a crime.

Part 4: Disclosure: when to label AI content

Label AI content when viewers could reasonably believe it’s real. A cartoon mascot is obviously illustrated. A photorealistic image of a “customer” or an “event” is not.

Platform rules: Major platforms, including YouTube, TikTok, Instagram, and Facebook, require creators to label realistic AI-generated or altered content, and they may add labels automatically when they detect it.

Content credentials: Some tools attach invisible content credentials (a standard called C2PA) that record that an image was AI-generated or edited. Don’t try to remove them.

Simple ways to disclose:

• A caption line: “Image created with AI.”
• The platform’s built-in “AI-generated” label setting
• A small credit line on printed materials

Disclosure isn’t an admission that you did something wrong. It builds trust, and trust is the most valuable thing a brand has.

Part 5: Never use images to deceive

Some uses of AI images are wrong even with a label, because the main message misleads people:

• Fake reviews or testimonials, including AI “customers” praising your product. The FTC’s rules ban fake reviews and allow penalties.
• Fake “before and after” photos that show results your product doesn’t deliver
• Product images that look better than the real thing, such as a bigger portion or a brighter color
• Fake events or crowds, like a “packed grand opening” that never happened
• Misleading news or political images, such as an altered photo of a real event. Many states regulate AI deepfakes in elections.

The test: “If someone found out how this image was made, would they feel tricked?” If yes, don’t publish it.

The pre-publish checklist

Run every AI image or video through these six questions before it goes public:

1. Inputs: Do I own or have permission to use every photo and reference I uploaded?
2. Originality: Does it avoid famous characters, logos, and copied artwork?
3. People: Does any real, identifiable person appear? Do I have their written permission?
4. Accuracy: Does it truthfully represent my product, service, place, or results?
5. Disclosure: Could viewers think it’s real? Is it labeled where needed?
6. Trust: Would anyone feel tricked if they learned how it was made?

A worked example: a gym’s ad campaign

Tyrone owns a neighborhood gym and wants AI images for a New Year campaign. His first ideas, and how the checklist changes them:

First idea	Problem	Fixed version
A famous athlete lifting at his gym	Likeness without permission	An original, clearly fictional athlete, labeled as AI
“Before and after” photos of a member	Fake results; deceptive	Real member photos, with the member’s written consent
A packed class with 40 people	His classes average 15; misleading	An energetic class of about 15 people
His logo on a pro team’s jersey	Trademark problem	His logo on a plain gym shirt
A photo of his real front desk staff, “improved” with AI	Altering employees without permission	Real photo, edited only for lighting, with staff approval

The final campaign is honest, legal, and still eye-catching. Two members tell him they joined because the photos looked like real people at a real gym.

Think deeper: the questions nobody has fully answered yet

The rules for AI images are still being written. That makes this one of the most interesting parts of the field.

Was it fair to train on artists’ work? Many image models learned from huge collections of online images, including copyrighted art, often without asking. Artists and companies have sued, saying that’s copying. AI companies argue it’s fair use. Courts haven’t fully settled it, and the answer may differ by country. Some companies now offer models trained only on licensed images, which matters for business use.

Who owns an AI image? The U.S. Copyright Office says protection depends on how much a human creatively controlled the result. A prompt alone generally hasn’t been enough. Substantial human editing, selecting, and arranging can count. Exactly where the line falls is still being tested.

How will we know what’s real? Two approaches are developing: hidden watermarks in AI images, and content credentials (a standard called C2PA) that record how a file was made and edited. Neither is foolproof, so trust still depends partly on creators being honest.

What happens to creative jobs? These tools are changing work for illustrators, photographers, and voice actors. Thoughtful creators credit people, pay them when they rely on their work, and don’t imitate specific living artists.

A simple guide while the rules settle: create in ways you’d be comfortable explaining to your audience, to the artists whose work trained these tools, and to the people in your images.

Questions to sit with:

• Human artists learn by studying other artists’ work. Is an AI learning from millions of images the same thing, or different? What would change your mind?
• If you spent an hour refining an AI image, does it feel like your work? What would make it feel more or less like yours?

You’ve finished AI Images & Video

You can now explain how image and video models work, write professional prompts, control style and light, edit and repurpose photos, direct short AI videos, keep a brand consistent with a prompt kit, and publish honestly and legally. That’s a stronger foundation than most people creating with AI today.

Complete the knowledge check to earn your AI Creator badge. If you’ve finished every course, you’ve earned the title R.I.C.K. Certified AI Operator. Go make something great, and make it honestly.