Endeavors AI Podcast
Who Actually Owns AI-Generated Content?
Entertainment & esports attorney Justin Jacobson on IP, copyright, and AI.
Guest: Justin M. Jacobson, Esq. — New York–based entertainment and esports attorney with nearly 15 years of practice at the crossroads of music, fashion, sports, and gaming. He focuses on intellectual property (copyrights and trademarks), authored a book on esports law, teaches at universities, and is a former podcast host.
Episode summary
Justin Jacobson has spent the better part of a decade at the intersection of music, fashion, sports, and gaming — and now AI is hitting his clients from every direction. In this episode he explains what's actually settled versus up-in-the-air in AI copyright law, why only human contributions get protection, the training-data lawsuits reshaping the field, the confidentiality traps catching lawyers, and the practical contract terms creators and entrepreneurs should build in before they publish anything AI-assisted.
Key takeaways
- Generative AI has taken media and entertainment by storm, and the central question is copyright: what rights exist in AI outputs, and who owns them.
- The current rule: courts and the Copyright Office protect only human contributions. AI-generated portions are essentially a free-for-all — which breaks a media business built on exclusive rights and licensing.
- To register an AI-assisted work, you must disclose what the human did and that AI was involved; only the non-AI parts get protection (e.g., your written script may be protected while the AI images aren't).
- Training data is the deeper concern: models scrape existing copyrighted works, so outputs can incorporate protected material — creating liability you may not discover until the rights-holder finds it. (Watch for media companies buying AI companies to scrape "legally.")
- Lawyers keep getting sanctioned for AI-hallucinated citations in briefs — and now courts can sanction both sides if opposing counsel fails to flag fake cases. Always fact-check.
- Confidentiality risk: platform owners can often access whatever you put in, which threatens attorney-client privilege — sanitize identifiable information.
- The "icing on the cake" framework: use AI to refine and finish work (a couple more supporting cases, polish), not to bake the whole cake — those who let it do the entire job get burned.
- Know each platform's strengths and weaknesses: AI excels at digesting and summarizing information, but doesn't understand nuance (e.g., it made contract terms "mutual" where they shouldn't be).
- Build protections into contracts with third parties (editors, designers, videographers): warranties, indemnification, and a requirement to disclose AI use, so you can hold them liable and know what's protectable and monetizable.
- The one question before shipping AI-assisted work: do you have the commercial rights? Commercial use (including monetized social posts) often requires a paid enterprise/business tier.
- Where IP law is heading: state, federal, and international action (the EU directive, deepfakes, name-image-likeness, AI impersonation), with big media-vs-AI lawsuits driving how courts follow.
Full transcript
Collin McKee
Welcome back to the Endeavors AI Podcast, where we talk about the intersection of artificial intelligence, law, business, and the new economy being built around all of it. Today's guest has spent the better part of a decade at that exact intersection, long before most people saw it coming. He's a New York–based entertainment and esports attorney who built his practice at the crossroads of music, fashion, sports, and gaming. He's also a professor, a former podcast host, and one of the sharper legal minds I've had the pleasure of speaking with. Justin Jacobson, welcome to the show.
Justin Jacobson
Thanks for having me. I'm glad to be here.
Collin McKee
Before we get into the meat of it, give folks a quick picture of how you started and how you ended up here — esports law isn't the most well-known field, and it's a big thing in your part of the world in New York City.
Justin Jacobson
I've been an attorney for almost fifteen years now. I started in the more traditional entertainment, music, sports, and fashion space, working with all kinds of creatives — musicians, songwriters, DJs, professional athletes, fashion designers — and the professionals supporting them: agents, managers, and businesses in those verticals. About a decade ago I noticed what was happening in esports and professional video gaming — Fortnite, Call of Duty, League of Legends — individuals competing for thousands, if not millions, of dollars at events and stadiums around the world: the Staples Center in LA, Madison Square Garden in New York, venues across the globe. I've been a lifelong gamer, so I always knew esports was huge in Asia and South Korea, and about a decade ago I saw it grow here in North America. It opened my eyes — businesses and talent making real money playing the games I grew up with, Street Fighter and Mortal Kombat and Tekken and Smash and Madden. I investigated the professional side, started working with agents on behalf of their players, familiarized myself with the talent, the teams, and the deal structures, and developed my own book of business. I wrote a book on esports law, which gave me an interesting way to present it, and I've been able to teach at different universities and build curriculum based on the book. It's wild to see how YouTube, Twitch, and TikTok content creators have become the new modern-day celebrities, making tens of millions of dollars.
Collin McKee
You're working with people making six and seven figures playing video games — something unheard of fifteen or twenty years ago. That's a whole different episode on its own. But let's talk about the focus of my podcast, AI, because AI is hitting your clients from every direction. Set the stage from where you sit, advising creators, athletes, and entrepreneurs — what has AI actually changed about all this IP and copyright?
Justin Jacobson
My practice really focuses on intellectual property — copyrights and trademarks — and how the entertainment and media verticals intersect. Anyone who's seen the news has seen AI used from movies to TV to music — there was a song almost nominated for a Grammy that was completely AI-made. It took the entertainment and media space by storm, all based on copyright usage and what rights exist in these outputs. With generative AI, you can create an image, a movie, or music from a prompt — "I want salsa with a reggae loop and a trumpet every other note" — and the AI creates it. So it's become interesting in my practice to see how it impacts creatives: making their jobs easier and more efficient, but also potentially replacing their jobs.
Collin McKee
It's that human touch that's the linchpin. If you can't prove you actually made it yourself, you might not own it — that could be ChatGPT's IP, or no one's.
Justin Jacobson
Or, the way the courts say, no one's IP — it's everyone's IP. If it's not exclusive to the creator, then anyone can use it. That's the dilemma. Right now the courts and the Copyright Office only protect the human contributions; anything the AI puts out is a free-for-all. That doesn't work in the media and entertainment business, which is based on exclusive rights and licensing of IP. If you don't have exclusive rights, you can't license them, which means you can't monetize them, and anyone could use it. Why spend money and time making a movie if anyone can then use it, sell it, and stream it?
Collin McKee
So what's actually settled law versus up in the air?
Justin Jacobson
We've had Court of Appeals decisions, denied by the Supreme Court, and the Copyright Office released an AI guide: only human contributions receive protection; anything created by AI doesn't. To get protection in an AI-created work, you submit an application that outlines the human contributions and indicates AI was involved, and they protect only the non-AI work. So if you use Midjourney to create images for a project but write the story or script yourself, the script is protected as a literary work, but the images might not be. If you just said "I want a sun setting over mountains" and the AI made it, I can probably take that image and sell posters of it, and you can't stop me. If you had taken that photo as the sun was setting, you could stop me — that would be copyright infringement. So the way you approach these things changes: if you create something using AI, you have to understand the limits of your rights and how that affects monetization. We all love culture and music, but you also want to put food on the table, so there has to be the ability to monetize what you put out.
Collin McKee
And that's not even the training-data question, which really concerns me for clients building products — what are you doing to train the LLM, what piece are you putting forth yourself, and you've got exposure built into the output before you even publish.
Justin Jacobson
Right. These programs are trained on existing works — the bots scrape Spotify, and if you type in Drake, Rihanna, Jay-Z, it listens to every song and produces something with elements of all of them. It's learning from copyrighted work, which is the basis of a lot of these lawsuits: "you're training on our work." It's far-fetched that it won't start apportioning things that are one-to-one or close. The potential liability: if you use something trained on existing works, it might incorporate something that belongs to someone else, and you might have no idea until that company finds you're using their background photo — now they've detected it and you're infringing. You say, "I just got this from Midjourney; I didn't know they'd take your sunset and change the mountain colors but keep everything else the same." That's why you see media companies buying AI companies — to allow legal scraping, so if an output mirrors something, they had a license to it anyway.
Collin McKee
You touched on this in our first conversation — lawyers specifically are getting in trouble around confidentiality obligations and AI tools. Speak on that.
Justin Jacobson
Every time I open LinkedIn there's another story about a lawyer getting in trouble for an AI brief, because the AI hallucinates answers it doesn't know and will double down if you question it. Lawyers are submitting briefs without checking the citations or doing the due diligence. When you submit something to the court, you're essentially signing under penalty that you did the work, that you're competent, and that it's true. If you put forth fake cases, courts are sanctioning lawyers — suspending them, fining them, making them pay the other side's fees. And now judges are sanctioning both parties if you don't point it out: if I see my opponent using fake AI cases and I don't bring it to the court's attention, I can be liable for not upholding my duty to the court. So it's now my job to check every citation they put forward. And you have unsophisticated people putting personally identifiable information, confidential material, and deal terms into these programs without realizing that the platform owners can access whatever you put in — which hurts attorney-client privilege.
Collin McKee
Right — it's inexcusable. It's not AI that's the problem; it's people being careless, lazy, and then signing their name to it.
Justin Jacobson
Yeah — "it was my associate" is the other excuse. That's not good enough. You have to check what the people who report to you do. That's part of being a supervisor.
Collin McKee
Let's flip it. You had a really good framework — the "icing on the cake." Walk people through it.
Justin Jacobson
People need to look at generative AI as a way to build and refine what you do, not to make it the whole crux of your work. That's where the people who submit fake briefs have problems. If you've done it right and you want two or three more cases, or a case in another jurisdiction — to add the icing on the cake — you can find those using the program. But if you say "generate this motion for summary judgment" and take it to do the whole job, you shouldn't be surprised when it does what you asked even if it's not the right answer. When people use it to bake the full cake, that's where the issues are. Use it for the finishing touches — the icing, the polish — and it can be used to perfection. You also need to know its strengths and weaknesses. Certain platforms can digest information and summarize effortlessly — those are home runs. "Here's the initial deal term, summarize the most important points" gives you a good starting reference. But "do my whole job" is where you have issues. There was a LinkedIn post this morning where the AI made everything mutual, and some of those things don't need to be mutual because the next paragraph addresses it — it doesn't understand how to do things properly. So if you use it to refine and do a better job of what you've already done, that's the sweet spot, at least now. I can't predict five or ten years out, but right now the people using it to do the bulk of their work are where you'll see more problems.
Collin McKee
And that applies to creators, athletes building a brand, and entrepreneurs using AI to build a product. So what should creators and entrepreneurs actually build into their contracts now? Give us practical stuff.
Justin Jacobson
Especially if you're using third-party service providers — editors, photographers, videographers who could use AI — build in warranties and indemnifications that protect you. If they don't have proper licensing, or their use infringes someone else's rights and now you're liable, you need the ability to turn around and hold them liable. Make sure anyone using AI for your work discloses what they're doing, so you understand what the AI did and what the human did — especially for something you want to protect and monetize. If you don't obligate the service provider to disclose, they don't have to. If a web or logo designer uses Midjourney and it rips off someone else's logo and you're now in trouble — or it makes an album cover and you get sued because you're a bigger artist — that's damaging, and you need the ability to recoup or hold them responsible. You also have to understand how your clients might use it. I have a musician who uses it to a degree; I tell him you don't own rights to certain contributions, so if you use a cool AI-generated saxophone, anyone could take it. Just be aware of where your rights extend and where they don't. You're advising on both sides.
Collin McKee
That almost answers my next question, but I'll ask anyway. For anyone who doesn't have an attorney yet — and I'd encourage any creator playing with AI to look Justin up — what's the one question they should ask before they ship anything AI-assisted?
Justin Jacobson
Make sure you have the commercial rights. There's a big difference between non-commercial and commercial rights. If you're going to use AI-generated work for a commercial purpose — monetization, making money, even ads on Instagram — as soon as you click the monetization button, it's a commercial use. A lot of these tools have enterprise or business models where you pay a larger fee to get commercial rights. If you use things outside those rights, you could have a problem and get an interesting letter from OpenAI for using things outside their terms.
Collin McKee
Well put. What's one thing any creator or entrepreneur should do this week when it comes to AI or IP?
Justin Jacobson
Understand the strengths and the legalities of what you're using. It's better to know than not to know. So many times people say, "Wow, I didn't realize — I thought I could do this." That's why you have to read and know. Ignorance of the law isn't a defense. Especially if you're using this to make money as a business or career, you need to know what you can and can't do. And if you don't understand, get someone who does, so you don't spend time and money promoting a song that's all AI-generated only to find that anyone can take it and no label or publisher will pay you, because anyone can use it.
Collin McKee
There's nothing that's really yours to own, unfortunately. You said you can't tell five or ten years out — it's hard to tell six months out — but where do you think IP law is headed in the next two years?
Justin Jacobson
You're starting to see legislation at the state, federal, and international levels — the EU directive, and various U.S. efforts pending. States have enacted things around deepfakes and name, image, and likeness, AI impersonations, to protect notable people from unauthorized usage, whether it's fake political ads — there was a whole scandal during the election with fake AI messages saying "you don't have to vote, the election's over" — or fake sexually graphic material of celebrities, fake nude photos that aren't real but look real. There's so much the legislature will involve itself in because of how impactful this is to all areas of society. We'll see how that leads, and then how the courts follow. Right now there are tons of lawsuits pending by the big media companies against the big AI companies, and when powerhouses square off, policies start to get implemented.
Collin McKee
Last question — where can people follow you and find out about your work?
Justin Jacobson
Definitely follow me online — jmjesq.com is my website, with contact info, articles, and information. I'm on Instagram and other socials as "Justin Jesq," and on LinkedIn. Always happy to connect — my DMs are open.
Collin McKee
Justin, this is exactly what I'd hoped — practical, real, no fluff. I appreciate you bringing your expertise and being straightforward. People really do need to be educated before they let these tools run wild. For everyone listening, links to Justin's work will be in the show notes. And if you have questions about building things with AI, hiring, and how to use it, that's what Endeavors AI is here for. Thank you for your time, Justin.
Justin Jacobson
Thank you so much for having me.
Explore our services
Get Started