Copyright, disclosure, and the murky ownership of AI-assisted creative work — and what marketers should know before they build a campaign on it
Some years ago, a creative director I admired told me that the best slogan she ever wrote came to her in the shower, and that she’d spent the following decade defending it from people who swore they’d thought of it first. She won every time, she said, because she could tell you exactly where the idea came from. She had drafts. She had a dated notebook. She had a story.
I’ve thought about her often lately, because a great many marketers are now producing work that has no shower, no notebook, and no story — only a prompt, a few seconds of whirring, and a result that looks, for all the world, like something a talented person made. It’s a remarkable convenience. It’s also, for anyone building a brand, a question that’s yet to be fully answered: who, exactly, owns what came out of the machine?
The honest answer, at least in the United States, is that sometimes nobody does. And that turns out to matter a great deal.
The Human in the Loop
American copyright law rests on a simple, almost old-fashioned premise: protection belongs to works of human authorship. The courts have been tested on this point and have held firm. In a case I followed with limited attention – well before I’d decided to form an AI-driven marketing company – a computer scientist sought to register an image he said was created entirely by his AI system, listing the machine itself as the author. The Copyright Office refused, the lower courts agreed, and this past March the Supreme Court declined to hear the case, leaving the human-authorship requirement firmly in place.
That ruling didn’t say that anything touched by AI is unprotectable. The Copyright Office has been careful on this point. Its guidance draws a line between using AI as a tool — the way a photographer uses a camera or a designer uses software — and letting AI supply the expressive choices itself. Human selection, arrangement, and meaningful modification of material can still earn protection. What generally doesn’t qualify is the raw output of a prompt, however clever the prompt may have been. Typing “a confident fox in a tailored suit, art deco style” is, in the Office’s view, more like giving instructions to a commissioned artist than like painting the fox yourself.
For the marketer, the practical consequence is sobering. If your new mascot, your packaging illustration, or the melody of your radio jingle came straight from a generator with little human shaping, you may have built a brand asset that a competitor could lawfully copy. You’d be in the peculiar position of paying to promote something you cannot fence in.
The Trademark Lifeline, and Its Limits
There’s some comfort here, and it comes from a different corner of the law. Trademark protection doesn’t depend on authorship; it depends on use in commerce and on whether consumers come to associate a mark with your business. A logo or a tagline generated by AI can, in principle, function as a trademark once it’s genuinely identifying your goods or services in the marketplace.
But trademark’s a narrower shield than many assume. It guards against confusion in the marketplace, not against every form of copying. A rival couldn’t slap your AI-designed fox on its own sneakers, but an artist might be free to reproduce the image in a context where no one would mistake it for your brand. Copyright and trademark are partners in a well-defended identity, and losing one of them leaves a gap worth taking seriously.
The Other Side of the Ledger
Ownership is only half the puzzle. The other half is whether the material you generated borrows too heavily from someone else’s work. These tools learned from vast oceans of existing images, text, and music, and the question of whether that training was lawful is still being argued in courtrooms on several continents. Those cases will take time to resolve, and marketers need not wait on their verdicts to act sensibly.
What matters day to day is the output. If you ask a generator for something “in the style of” a famous illustrator, or for a song that sounds just like a recent hit, you are steering the machine toward the very thing that can land a brand in trouble. Style itself is generally not protected, but a close resemblance to a specific, recognizable work can be. A jingle that echoes a chart-topper too faithfully is no safer for having been composed by an algorithm. If anything, it is riskier, because no one on your team can testify to having created it independently. My old colleague’s dated notebook was her best defense; a prompt history is a poor substitute.
Disclosure Is Becoming the Default
Meanwhile, a separate current is reshaping the conversation: the expectation that audiences be told when they’re looking at machine-made material. In the European Union, the transparency provisions of the AI Act took effect in August of this year, requiring that certain AI-generated or manipulated content — deepfakes most prominently — be identified as such. Several American states have enacted their own rules on synthetic media in political advertising, and the major platforms have introduced labels of their own for realistic AI imagery.
I’d gently suggest that marketers treat these rules less as a burden than as a preview. Customers are becoming fluent in spotting synthetic content, and they tend to forgive what is disclosed far more readily than what is discovered. A brand that’s quietly found to have used AI-generated “customers” in a testimonial video, or a fabricated photograph of a product that doesn’t quite exist, pays a price in trust that no efficiency gain can recover. A brand that says, plainly and without fuss, “this illustration was created with AI tools,” usually pays nothing at all.
A Practical Checklist
None of this argues for abandoning the tools. They’re genuinely useful for brainstorming, for drafting, for exploring visual directions at a speed that would have astonished the creative departments of my early career. The question isn’t whether to use them, but where in the process to let them stand. A few habits go a long way:
- Use AI early, not last. Let it generate rough concepts and options, then have human designers, writers, and composers shape the final asset. The more meaningful human authorship in the finished work, the stronger your claim to it.
- Keep the paper trail. Save drafts, revisions, and notes on what your team changed and why. If you ever need to register a copyright or defend an asset, that record is your story.
- Read the terms of service. Generator licenses differ on who owns outputs, how they may be used commercially, and whether the vendor offers any indemnity against infringement claims. Those details belong in the conversation before a campaign launches, not after.
- Avoid prompting toward a specific living artist, a recognizable song, or a competitor’s look. Imitation invites exactly the comparison you hope to avoid.
- Be especially careful with core brand assets. A social media graphic that runs for a week is low stakes. A logo, mascot, or sonic signature you intend to use for twenty years deserves human authorship and a conversation with an intellectual property attorney.
- Decide on a disclosure policy and apply it consistently. Clarity is easier to maintain than a patchwork of case-by-case judgments.
The Story Still Matters
It strikes me that the deepest lesson here isn’t really a legal one. The law is simply catching up to something good marketers have always known: a brand’s value lies in its distinctiveness, and distinctiveness comes from human intention. The courts protect human authorship because human authorship is where the meaning lives — the choices, the judgment, the small, surprising decision that makes a slogan unforgettable.
My friend with the shower slogan could defend her work because she knew where it came from. The marketers who’ll thrive in this new era are the ones who can still say the same, even when a machine helped them along the way. Use the tools generously. Just make sure that, when someone asks who owns the words, you have a good answer — and a story to go with it.
This column offers general information, not legal advice. Laws governing AI-generated content are evolving quickly and vary by jurisdiction; consult a qualified attorney before relying on any specific asset.