Volume Out, "Fine Tuning" In for GenAI Video Licensing
Focused Video Content, Not Bulk, Is What GenAI Developers Want Now
Welcome to my Monday morning “brAIn” dump — the first under our new, shall we say, “regulatory un-friendly” AI regime. As always, first, it’s the“mAIn event” — my feature about how generative AI video content licensing has significantly evolved over the past months. Then, the “mosAIc” — this time, a timely story about AI and the Oscars. Finally, the “AI Litigation Tracker” — updates on key GenAI/media cases by law firm McKool Smith (you can also check out the “Tracker” here via this link).
But First …
A picture says $500 billion words …
That’s an aid to OpenAI CEO Sam Altman and his Oracle and SoftBank compatriots handing over the keys of Silicon Valley to the new Sher-AI-ff in town (as they announce their $500 billion mega-AI infrastructure joint venture and end the promise of at least some basic AI guardrails coming out of D.C.). Yes, we’re now squarely in Wild West AI Accelerationist land. State governments — and our independent judiciary — are where copyright, NILV, privacy & other content-related guardrails will be defined (including infringement v. fair use). Buckle up, it’s gonna be a bumpy ride.
P.S. And as for artificial general intelligence (AGI) — and the potential existential risks posed by it (that even Trump’s VP Elon warned about in a galaxy not so far away) — well, no need to worry. Altman — whom I predict will announce OpenAI achieving AGI this year — tells us he has it all under control. So, why worry? Right?
I. The mAIn event - Hey Media Companies, It’s All About Video “Fine Tuning” Now
NOTE: You can listen to a podcast discussion of this article, generated by Google NotebookLLM, via this link or by clicking on this button below.
Intro
There was a time when “bulk” was the metric that mattered most to Big Tech for video content (“data”) needed for GenAI training purposes. In fact, those times were just months ago. “The more, the better!” was Silicon Valley’s mantra. But those times are over, and they aren’t coming back. Licensing’s life-cycle has moved on.
Welcome to the brave new world of video licensing — GenAI “fine tuning” style.
What Is Fine Tuning?
“Fine tuning” refers to the concept of AI developers optimizing their already-developed LLMs — many that likely have been significantly trained on copyrighted works — by focusing on filling important “holes” missing from the content they’ve already scraped or licensed. Video licensing deals fall squarely in this “fine-tuning” zone right now. It’s all about specific “asks” by generative AI research teams — “asks” that continuously change over time.
Generally speaking, those GenAI researchers already have the volume of video “data” they need (“data” is what tech companies call it, but it’s prized “content” to the creative community). In the words of one key “data” licensing tech exec, it’s no longer about content “tonnage.” “Specialty video libraries” are all the rage. And different GenAI research teams have different needs.
I know this, because I’m right in the middle of these discussions to find, define and negotiate fair “win win” licensing deals.
Dynamic Video Is All The Rage
Static video “talking heads” (e.g., news reporters) are no longer a priority for most tech giants. Now it’s mostly about dynamic content — motion and movement. High resolution video (1080p and above) and robust meta-data are essential too, not to mention “contextual data” that fills in other training gaps. Contextual data refers to “b-roll,” “cutting room floor,” and other video that has never seen the light of day but gives other video context. If a video library meets these parameters and is of sufficient size (typically 10,000-25,000 hours minimum), then there’s significant licensing interest. Right now, in fact.
So size still matters for GenAI developers. But it must be the “right” kind of size. The more of that coveted focused, fine-tuning friendly “specialty” video content, the better. Overall per minute licensing terms (for training purposes) grow as available content grows, and those economics grow even more if the content licensor also agrees to license so-called “display rights” — which give the GenAI licensor the right to display the content itself in GenAI outputs.
Of course, it’s fair to ask how GenAI developers have reached the “fine tuning” stage for their LLMs when no GenAI video licensing deals have been publicly announced — especially since OpenAI launched Sora nearly one year ago (and most others have followed suit). That video content on which Sora and others are based — a ton of it — obviously had to come from somewhere in order to train their foundational LLMs. Any guesses anyone?
Litigation “Friction” Holds Back GenAI Adoption
No surprise, then, that media companies are now litigating over 40 copyright infringement and related cases in the courts. I’ve written extensively on the subject (here’s one of my recent relevant articles), and you know how I feel about it all from both a legal and ethical perspective. Unconsented and uncompensated taking is no “fair use.”
But, for now, let’s set aside that fundament issue that now sits in multiple federal courts. Let’s get back to the central content/”data” licensing issue at hand.
Now is the time for content licensing discussions, negotiations, and agreements that fairly compensate creators and rights-holders. After all, their content is the rocket fuel that enables LLMs to take off, optimize and commercialize at scale. And it’s not just in the interests of the creative community. Litigation in the courts — together with Big Media’s fear of liability for using generative AI tools and the creative community blowback that would come with it — are friction that holds back the kind of mass GenAI adoption Silicon Valley craves. The best way to eradicate that friction is by eradicating litigation — i.e., by entering into fair “win win” solutions that reflect the essential value content adds to the AI equation.
The Great “GenAI Unlock” Comes from Win-Win Licensing
The great “generative AI unlock” will only come when the copyright litigation goes. And that’s a long, long, long time for now (I started my career as a federal district court chief judge’s law clerk, so I’ve had a front row seat to how slowly the appellate process goes). So, it’s in everyone’s interests for fair video content licensing deals right now.
(For those of you interested in learning more or exploring licensing opportunities, reach out to me at peter@creativemedia.biz).
II. The mosAIc — It’s Oscar Season & Noms Are In!
This Oscar season, all the buzz surrounds the multiple Oscar nominations for “The Brutalist” and “Emilia Perez,” both of which extensively used AI tools to clone voices. The creative community was mixed and conflicted about it all — with some bemoaning AI being used to give “The Brutalist”’s lead Adrian Brody a fluent Hungarian voice (which is particularly poignant to me, since I am 100% Hungarian and could absolutely distinguish between Brody actually speaking Hungarian from when AI did the heavy lifting for him).
In any event, AI’s ever-greater use in Hollywood is making quite a statement, as is evident in this recent feature story in CNBC (in which I am extensively quoted with my own views). Does AI enhance art? Diminish it? A mixed bag?
Talk amongst yourselves …
Follow me on BlueSky via this link.
You can also continue to follow my longer daily posts on LinkedIn via this link.
III. AI Litigation Tracker: Updates on Key Generative AI/Media Cases (by McKool Smith)
Partner Avery Williams and the team at McKool Smith (named “Plaintiff IP Firm of the Year” by The National Law Journal) lay out the facts of — and latest critical developments in — the key generative AI/media litigation cases listed below. All those detailed updates can be accessed via this link to the “AI Litigation Tracker”.
You can also listen to a podcast discussion of the latest updates (as of January 25th, 2025) via this link or by clicking on the button below.
You can also find this episode on Peter Csathy’s “the brAIn” and “Fearless Media” podcasts which are available on all major podcast platforms.
(1) Raw Story Media v. OpenAI
(2) The Center for Investigative Reporting v. OpenAI
(3) Dow Jones, et al. v. Perplexity AI
(4) The New York Times v. Microsoft & OpenAI
(5) Sarah Silverman v. OpenAI (class action)
(6) Sarah Silverman, et al. v. Meta (class action)
(7) UMG Recordings v. Suno
(8) UMG Recordings v. Uncharted Labs (d/b/a Udio)
(9) Getty Images v. Stability AI and Midjourney
(10) Universal Music Group, et al. v. Anthropic
(11) Sarah Anderson v. Stability AI
(12) Authors Guild et al. v. OpenAI
NOTE: Go to the “AI Litigation Tracker” tab at the top of “the brAIn” website for the full discussions and analyses of these and other key generative AI/media litigations. And reach out to me, Peter Csathy (peter@creativemedia.biz), if you would like to be connected to McKool Smith) to discuss these and other legal and litigation issues. I’ll make the introduction.
About My Firm Creative Media
My firm and I specialize in market-defining strategy and content licensing for generative AI, breakthrough business development and M&A, and cost-effective legal services in the worlds of media, entertainment, AI and tech. We develop game-changing strategic opportunities and leverage our uniquely deep relationships to reach key decision-makers and influencers in record time to execute. Not just talk.
Among other things, we represent media companies for generative AI content licensing, with deep relationships and market insights and intelligence second to none. Reach out to Peter Csathy at peter@creativemedia.biz if you’d like to explore working with us.
Send your feedback to me and my newsletter via peter@creativemedia.biz.