Is Gen AI Training on Copyrighted Material Legal? 🤔
Two major court rulings just nudged us a step closer to clarity on one of the most pressing questions in advertising and content creation: Is it legal to train Gen AI on copyrighted material?
In short? Sometimes. In Kadrey v. Meta, a judge ruled that training on copyrighted books could qualify as fair use, at least under the specific facts of the case. In Bartz v. Anthropic, a separate judge came to a similar conclusion: if the content was lawfully acquired and not directly competing with the original, it’s likely okay. However, both cases also flagged that retaining unauthorized copies could still be problematic, and that battle isn’t over yet.
Why This Matters for Advertisers and Content Owners.
For creators, the takeaway is this: while courts may be softening on Gen AI training, the line between “fair use” and “unauthorized use” is still blurry. And public trust? Even blurrier.
A recent example: A major video platform quietly trained its Gen AI models on user-uploaded videos. Outrage followed. People don’t like feeling like their creative work is being mined without consent: legal or not.
Even Coca-Cola’s Gen AI holiday ad, which recycled beloved imagery, backfired with many calling it “soulless.” It’s not just about legality. It’s about human connection, and whether your audience can feel it.
At Qortex, we live and breathe the future of intelligent video. But here’s the truth:
Just because something is legal doesn’t mean it’s good.
The platforms may be winning early rulings, but brands and creators should tread carefully.
Use Gen AI to enhance creativity, not replace it.
Build with tools that:
•Keep your data yours
•Don’t train on your inputs unless you say so
•Offer indemnity and transparency
•Allow for meaningful human input
Because the audience can tell when something’s missing—and it’s usually you.
Let Gen AI do the heavy lifting, but don’t let it strip your message of soul!
Talk soon,
Zack Rosenberg Get in Touch about engineering outcomes from video!
From My Feed: Most-Impactful Post of the Week
This week’s top post is a love letter to local advertising—especially the bold, memorable, sometimes unhinged world of local law firm marketing. From shotgun-starring car dealers to salon ads with toddler actors, nothing beats the creative freedom of small, scrappy, community-first brands.
Local lawyers know something most marketers forget: you don’t win by being cheap, you win by being unforgettable. They spend big, not for immediate ROI, but to lock in long-term mindshare. Whether you need them now or 10 years from now, you’ll remember their jingle, their number, or even their breakup musical (yes, really).
Tired of chasing cookies or building broad audience profiles that miss the mark? Qortex’s latest blog dives into Inferred Audiences—our AI-driven way to match your message to real-time viewer mindset based on video context, sentiment, and engagement signals. No personal data. Just precision targeting rooted in what’s actually happening on screen.
Why it matters:
Traditional targeting relies on who someone is. Inferred Audiences focus on what someone cares about in the moment—leading to higher engagement, better ad recall, and more meaningful results.
Try It Yourself:
Explore the blog and see how Inferred Audience Finder can help you meet your viewers in the moments that matter most.
This is going to be interesting to watch over the next few months. The ruling should incentivize the use of higher-quality, legally obtained content for training, and I would expect the innovation in this space to pick up. That should at least help the content creators monetize their work more. I agree 100% that there AI-generated content usually lacks the spark of reality and humanity that we all crave and can sense when it is absent. But I suspect the threshold will be lower for some types of content and "good enough" may be fine for many people in certain scenarios. The trick--and where I see the future litigation coming--is when companies monetize that work and it competes with the work of original creators. The ruling fell short of addressing that, and "fair use" clearly doesn't allow for that.