Hey Marketing Bestie,

I married into a family of lawyers and it's handy. Long before AI could help me, Davey could (my father-in-law, who will want me to tell you he's never given me legal advice and has just heard me out in the hypothetical sense).

Speaking of legal...

Was this email forwarded to you?

Subscribe Here

6 AI Marketing Questions Your Legal Team Needs to Answer

Ok so let's set the scene.

You've got an AI agent writing your ad copy, another one picking your audience segments, and a third one you forgot you even signed up for that's "optimizing" your email sends at 2am.

Cool. Efficient. Very "future of marketing."

Also… does anyone actually know what any of these tools are doing with customer data right now? Anyone? Bueller?

Yeah. That's the problem. Le sigh.

Regulators noticed the same thing we did. AI marketing went from "cute chatbot" to "fully autonomous budget-and-targeting machine" in about 18 months, and now the paperwork is catching up. The EU AI Act's high-risk AI system obligations take full effect this month, which, unless you've been living under a rock (no judgment, sounds peaceful), is basically now.

So here's your marketer-friendly, non-lawyer, "please don't get in trouble" rundown. 6 things.

6!

1️⃣. Know what "high-risk" actually means for YOU

Not every AI tool is regulated the same way. But if your AI is making decisions about who sees an ad, what price someone gets, or whether someone qualifies for something (credit, a job posting, insurance), that's the category regulators care about most.

Here's what I would do: Ask legal: "Does anything we use for targeting, personalization, or pricing count as high-risk under the frameworks that apply to us?"

Don't guess. Ask.

2️⃣. Written by AI" might need to be a label, not a secret

Disclosure rules for AI-generated content are catching up fast. Synthetic reviews, AI spokespeople, AI-written testimonials. Regulators want to know when a human isn't actually the one talking. Consumers kind of want to know too, turns out.

Here's what I would do: Ask legal where the line is on disclosing AI-generated content, especially anything that looks or sounds like a real person's opinion.

3️⃣. Your AI copy tool is confidently, beautifully, sometimes wrong

Generative AI can hallucinate product specs and claims that were never true. Says it with SO much confidence too. Zero hesitation. A LinkedIn thought leader could learn from this energy, honestly.

Here's what I would do: Any AI-generated claim about pricing, specs, health, or performance gets a human fact-check before it goes live.

Every time. No exceptions for "it sounded right."

4️⃣. Opt-out rights are becoming a whole thing

State-level laws (California's DELETE Act is one example) are giving consumers more say over automated profiling. Translation: people can increasingly say "please stop building a shadow profile of me," and companies have to actually listen.

Here's what I would do: Ask legal what your opt-out and data-deletion process actually looks like end to end. Not just "we have a form somewhere."

5️⃣. "The AI did it" is not a defense

This one's less legal-code and more vibes-based truth: regulators and the FTC are looking at the marketer, not the model, when an AI-driven ad makes a false claim. The tool doesn't take the meeting when it goes wrong. You do.

Here's what I would do: Build a lightweight approval step for anything AI touches before it hits a live audience.

Boring? A little.
Better than the alternative? Absolutely.

6️⃣. Read the terms before your customer data becomes someone else's training set

A lot of AI tools quietly reserve the right to train on whatever you feed them. That includes customer emails, chat transcripts, campaign data, all of it, unless you specifically opted out. Convenient for the vendor. Less convenient for the customer who never agreed to that.

Here's what I would do:: Ask legal to actually read the data usage terms on your AI tools, not just the pricing page.

Find out if there's an opt-out for training data, and use it.

ENDING IT HERE...

None of this means panic and unplug every AI tool in your stack (please don't, your Q3 roadmap would never recover). It just means "we didn't know" stops being a great answer.

Go talk to your legal team. Be nice as you go!

MEME OF THE WEEK

Sponsored by Community

Really great Marketing is about listening more than you talk.
That goes DOUBLE when you're texting customers. TRIPLE during the holidays. 📲🎄🎁
This free SMS playbook shows you how - before, during, and after Black Friday.

Inside:
--> 5 things to do BEFORE October 1st to set yourself up
--> Text prompts to steal
--> A week-by-week texting timeline through Cyber Monday

If real, 2-way text conversations feel like fantasy, download the playbook and make it real now - before November.

Get the Holiday SMS Playbook

🎙 TUNE IN

Podcast to listen to this week: 

How to Manage AI Agents Like Employees

What you will learn:
🎙: Why treating your AI agent like an employee generates $13M in productivity
🎙: The mistake everyone makes when they build digital workers
🎙: How entry-level humans become managers when they start coaching AI

Listen Here

JUST FOR FUN
🤩: Brand of the week: Delta
😛: Favorite meal I ate: Tatte
🎵: Favorite song this week: Asleep Talking
💍: What I bought my wife: Yellow button down.

In 24-hours my son will try his 1st California burrito. Is there anything better than your kid finding a new favorite food? He currently loves Ramen most, followed by pancakes lol.

Your friend,
Daniel