Marketing Teams Got Hooked on AI. The Law Isn't Cutting Anyone Slack.

A piece of news crossed my feed recently that made my stomach drop.

In late 2025, Figma got sued. The plaintiff alleged that the company took customer design files — images, text, the works — and used them to train its own generative AI models. No warning. No permission.

Think about how many marketing teams have their lifeblood sitting in Figma. When the story broke, it hit people: the stuff I build on your platform gets turned around and fed into your AI training pipeline? What am I, then?

And that was just the beginning.

Around the same time, Valentino released a luxury ad campaign made with AI. Surreal visuals, dazzling technique. The result? The internet tore it apart. People called it "tacky." They said it threw away decades of craftsmanship the brand had painstakingly built. Coca-Cola tried AI for holiday ads, too — and got blasted by consumers and the creative industry alike.

Do these brands lack money? Lack technology? They don't lack either. But AI taught them a lesson: just because the tech works doesn't mean you can afford the fallout.

The Law Isn't Going to Coddle AI

A lot of people harbor an illusion: AI is new, so the law hasn't caught up yet, right?

Quite the opposite.

Copyright law, data privacy law, consumer protection law, false advertising law — not one of them says "AI, you get special treatment." Regulators and courts have stated repeatedly: when things go wrong, we're not coming after the AI tool. We're coming after the company that used it.

Saying "the AI generated that on its own" carries about as much weight as telling a cop "the car crashed itself."

So where are the landmines? I picked the five that marketing folks are most likely to step on.

Risk #1: Who Owns the Copyright?

How did AI learn to write copy and draw illustrations?

By "eating" data. Massive quantities of text, images, audio, video. Are there copyrighted works in that training data? Almost certainly. And could the output collide with some creator's original work? Absolutely possible.

That creates a deeply awkward situation.

As of early 2026, U.S. courts have been handing down contradictory rulings on whether training AI on copyrighted material counts as fair use. Some judges say the transformation is substantial enough to qualify. Others say that if the tool directly eats a creator's lunch, it doesn't qualify. Meanwhile, the U.S. Copyright Office has made its position clear: works generated purely by AI are not eligible for copyright protection.

Let that sink in.

You generate a poster with AI and publish it — you might be infringing someone else's copyright. You try to claim rights over that same poster? Sorry, it doesn't even belong to you.

Boxed in on both sides.

What to do? Three things. First, before signing any contract, investigate where the AI vendor's training data comes from and whether they have proper licensing. Second, bake IP (intellectual property) compliance indemnification and warranty clauses into the contract. Third, run a pre-publish review and check for obvious traces of third-party assets.

Risk #2: User Data — the Line You Don't Cross

AI-driven marketing runs on data. Behavioral tracking, predictive modeling, automated decision-making. The more you feed it, the more accurately it predicts.

But more data also means more risk.

The EU's GDPR (General Data Protection Regulation), California's CCPA (California Consumer Privacy Act), and a steady stream of new regulations from other countries are all watching. The FTC (Federal Trade Commission) in particular has been hammering the message: use data opaquely, engage in fuzzy profiling, touch sensitive information — especially anything involving children — and that's an unfair and deceptive business practice.

The FTC said something I think nails it: "algorithmic opacity" is not a get-out-of-jail-free card.

AI systems can infer far more about a consumer than the user ever voluntarily provided. The more accurately you target, the tighter the privacy tension gets.

How to put this into practice? Work with legal and IT to conduct a data protection impact assessment before launch. Confirm you have a lawful basis for profiling and targeted advertising. Commit to data minimization, maintain transparency, and keep a human in the loop as a backstop.

Risk #3: AI-Generated Ads Still Have to Play by the Rules

Some people figure that if they're using AI to make ads, slapping on an "AI-generated" label makes everything fine.

It doesn't.

A label doesn't fix anything. If your AI-generated image exaggerates a product's effects, fabricates user testimonials, or implies unverified claims — especially in heavily regulated sectors like healthcare and finance — you're still on the hook. Regulators have been crystal clear: when AI-generated content in an ad goes wrong, the responsible party is the advertiser, not the AI tool.

So AI-generated marketing materials should go through the same review process as anything else. However you managed human-created content before, manage AI-created content the same way. Equal treatment.

Also assess this: if you don't disclose that AI was involved in the creative process, will consumers be misled? If the answer is yes, don't hide it.

Risk #4: Synthetic Media and Deepfake Risks

Generative AI has dragged the barrier to making deepfakes down to the floor. A convincingly fake audio clip, a flawless video — you can produce one in a matter of hours.

What does that mean in a marketing context?

If you use AI to generate a realistic-looking "user" to endorse your product, and that person doesn't exist, you may have committed false endorsement. If you synthesize a real person's likeness or voice without their consent, you could face defamation or right-of-publicity claims.

And this isn't just a U.S. problem. EU member states, India, China, and Japan are all legislating against deepfakes. In 2026, the regulatory screws are tightening further worldwide.

The red line is clear: no explicit consent, no recognizable real people in your AI marketing content. Period. Set up content review and escalation protocols, and make sure everyone on the team knows: defamation, harassment, impersonation — the legal consequences are severe.

Risk #5: Bias and Fairness

Finally, one that's easy to overlook.

Highly personalized AI systems can quietly amplify discrimination. You ask it to do precision targeting, and it might "precisely" exclude certain demographics. You ask it to generate content, and it might reproduce and reinforce stereotypes. And that's before you get to the people who deliberately weaponize AI with dark patterns to manipulate user decisions.

The signal from regulators is: even without AI-specific legislation on the books, existing consumer protection authority is enough to come after you.

But here's the reality — reputation collapses first, fines arrive second. Once an ethical issue goes viral, legal exposure is just a step behind.

The wall between ethical risk and legal risk is getting thinner by the day.

So build an internal AI governance framework. Run fairness and bias testing on your targeted advertising and personalization systems. Give consumers clear disclosure and accessible complaint channels. Doing these things now is an insurance policy for your brand.

Final Thoughts

AI is genuinely reshaping marketing. I believe that.

But it's reshaping marketing inside a legal environment that's growing more mature and more aggressive by the day. Copyright lawsuits, privacy enforcement, false advertising scrutiny, synthetic media regulation — every trend points to the same conclusion: AI risk is now a core governance issue for marketing.

Companies that treat AI purely as a productivity tool will probably learn their lesson the hard way — and at brutal cost.

The companies that front-load legal review, contractual safeguards, and ethical oversight into every stage of their AI marketing pipeline will be the ones who actually reap the technology's rewards.

Instead of becoming the next cautionary tale.

That's why I wrote all this down. I hope you won't be the one in it.