"Sora's" text-to-video shocked Hollywood. But generative AI can bring Lego-like artist building blocks to life for unlimited creative possibilities
The Optimist's case: creators can give their audiences basic creative elements that they can flesh out in the medium and format they choose
Happy Monday! Time for your weekly AI-infused media and entertainment brAIn dump. I’m here to give it to you straight (but also entertaining!) about the promise and perils of generative AI in the world of media and entertainment, together with insights, updates and “news you can use.” These are my perspectives, of course, and I welcome debate (and will feature other points-of-view) because things move so fast and perspectives likely will evolve. So here’s how we do this!
First, “the trAIler” - the key generative AI headlines from last week and preview for this week. Next, “the mAIn event” - my feature story for the week. Then, “the cocktAIl” - my curated mixology of AI events and helpful resources. Finally, “the AI legal case tracker” - updates on the most important generative AI-related copyright infringement litigations (because they impact everyone in media and entertainment).
I. the trAIler - 10 AI “quick hit” headlines and previews for the week (you know, like a movie trailer!)
(1) OpenAI’s “Sora” movie magic, or is it more like movie tragic? It all depends on perspective, as the ChatGPT creator announced its new photorealistic “text to video” generative AI platform (and Hollywood understandably gasped and grappled with its implications). Will largely AI generated Hollywood feature films be coming to a screen near you? My Magic 8 Ball says “signs point to yes.” But how soon, we don’t know. Doesn’t mean human filmmaking will ever disappear (it won’t - here’s why). But many current jobs will, even as new categories of jobs (like creative prompt engineering) are created.
(2) Meanwhile, movie mogul Tyler Perry blames “Sora” for his decision to halt his $800 million studio expansion. Or is his decision just a convenient excuse to stop the flow of cash? Either way, Perry’s “move” opens the door for others to follow. Another blow to traditional Hollywood jobs. It’s a huge story.
(3) Nvidia, take a bow! The OG generative AI chipmaker killed it with its quarterly earnings report last week. It added $272 billion to its market cap in just one day - the largest stock market gain ever! CEO Jensen Huang put it this way as his stock popped: “Accelerated computing and generative AI have hit the tipping point. Demand is surging worldwide across companies, industries and nations.” Is it too late to jump on the Nvidia train and invest? I invested 1 year ago (and am thinking about adding more). We are still only in the first inning of this generative AI game.
(4) Meanwhile, OpenAI CEO Sam Altman apparently doesn’t have enough on his plate (certainly not generative AI “trust and safety” apparently). Now he’s raising $5-$7 TRILLION (yes, you read that right, with a capital “T”) to take on Nvidia in the great generative AI chip wars. Meanwhile, Softbank is raising a fraction of that - “only” $100 billion - for its own Nvidia “killer” of his own. Good luck with that!
(5) Speaking of Sam AI-tman, OpenAI is now valued at over $80 billion. One year ago, its valuation was $29 billion. Many believed back then that it was over-valued. I didn’t (and am on record for saying that at the time).
(6) The New York Times is getting jiggy with it! If you can’t beat ‘em (although it’s trying with its case against OpenAI that I track below), then join ‘em. The Times is plotting to launch its own generative AI ad tool.
(7) Adobe is telling its generative AI story too, in pdf format of course. First, Adobe launched its “Firefly” text-to-image tool. Now it’s launching an AI assistant for Acrobat and Reader.
(8) It’s a big world out there, it’s not just U.S. copyright law. Case in point, India - which just recently mandated that AI developers must get consent to use copyrighted materials for generative AI training.
(9) Not to be outdone, the European Union is finalizing the text of its AI Act, which also would require AI developers to get permission for the use of copyrighted materials in their training.
(10) In other words, Google, you too are treading on very precarious ground. YouTube’s generative AI platform “Dream Track” creates music for use in YouTube shorts, but is trained on copyrighted recordings. That can’t be right!
II. the mAIn event - the optimist’s view of generative AI’s creative potential (and monetizing possibilities for artists): Lego-like creative building blocks for fans to unlock unlimited creative possibilities
Just over one week ago, ChatGPT creator OpenAI shocked Hollywood and the entire creative community yet again when it announced its new “Sora” text-to-video generative AI model. Sora boasts photorealistic video and promises to ultimately deliver full high-definition cinematic quality video, causing many to wonder what roles film schools will play in the future.
It’s just the latest generative AI tool that any user — no matter how versed they are in the arts — can use to unleash previously unimaginable and unattainable creative possibilities. Its power is both democratizing and daunting, causing many in the creative community to understandably fear the shedding of traditional industry jobs at its artificial hands. Tyler Perry, in fact (as mentioned above), just halted his $800 million studio expansion and blamed his decision on Sora’s “mind-blowing” power.
The generative AI optimists’ view
But tech optimists see things differently and believe that generative AI’s power promises to create entirely new categories of industry jobs – creative prompt engineering being one of them (the human skill to “dial in” the perfect story or film with text or voice). Even more transformational is generative AI’s anticipated power in the hands of storytellers. No longer constrained by traditional fixed storytelling “packaged” formats like the novel, audiobooks and podcasts, video shorts, video series and films, creatives will be able to build basic frameworks for stories on top of which their audiences can build, refine and package themselves.
Think of it this way. An artist essentially creates sufficiently fleshed out — and hence, copyrightable — Lego-like story building blocks and makes them available to audiences to create their own personalized creative works, rather than just the fixed one that the artist puts out into the world. It’s kind of like kids using Legos to build endless creations on top of Lego’s famous foundational green Lego baseplate. The creative possibilities are endless.
Imagine a scenario in which a writer lays out a basic skeletal story-scape, populates it with one or more “must use” main characters, and then makes it available to the world. That basic foundational story package becomes their Lego-like “baseplate.” Audiences, attracted to that story or the creator behind it, can then use generative AI to choose their own desired format in which it is told (short story, novel, audiobook, animated video, video game, episodic series, Hollywood-style film), the story’s setting, its additional characters, and its overall more specific plot points and story arc. Consider it a menu of infinite story possibilities.
Grimes and endless derivative works
This means endless derivative works (and derivative works on top of those derivative works) that are endlessly monetizable. Blockchain technology perhaps can finally be that long-promised great creative hope and mechanism to track and pay out the artist, after the artist has first set the basic rules of the game that all users must follow — including compensation and distribution terms.
Artists can, at least conceptually, also define other essential creative guardrails to preserve fidelity to their vision — such as prohibiting the use of guns and violence, false, defamatory or overtly political content. The entertainment industry is already deeply familiar with those types of restrictions that are frequently found in content licensing arrangements.
We already see this happening in the music industry with innovative artist Grimes leading the way. Grimes has always been tech-forward, well before she wed former Tony Stark (now Shark) Elon Musk. Less than six months after OpenAI’s ChatGPT rocked our worlds in late November 2022, Grimes launched her own entirely new kind of music platform Elf.Tech that made her voice and music “stems” (essentially music’s Lego building blocks) freely available to her fans to create their own Grimes DNA-infused works. Grimes’ only condition to her fans was that if any of their derivative works became commercially successful, she would share 50% of its royalties.
Grimes, in other words, created an entirely new kind of creative platform and business model for musicians that both delights her fans and potentially deepens her pockets. It also keeps Grimes’ “brand” alive on an ongoing basis, rather than just every now and then when she releases new music. Generative AI optimists also believe that these fan-created derivative works also complement, rather than cannibalize, Grimes’ own music market. Her fans still anxiously await her new music, precisely because they want to experience her unique human personality and latest complete, authentic and wholly original creative vision. It’s not an either/or proposition.
In the optimist’s mind, there is no reason that other forms of creativity — including written, audio and visual works — can’t follow that same path. Take horror master Stephen King who endlessly churns out fully fleshed novels that thrill. Now King can augment those published novels — which are fixed with his singular vision — by publishing unlimited Lego-like story “blocks” that flow from his restless mind, but for which he has no time to fully flesh out. With OpenAI’s Sora or other forms of generative AI, now King can give those Legos to his insatiable audience so that they can create their own horror franchises - and he, as the creator of those blocks, would share in the success and monetization of it all. King would also continue to monetize his own fully written new standalone works, of course.
Viewed in this optimistic light, generative AI’s essentially infinite creative possibilities exist for all forms of media. Taking the King example further, his audience could decide whether to generate their own King-based story in written, audio or visual form — which takes us back to OpenAI’s Sora and considers it in a more positive creative light.
Generative AI’s threat is real, but fear is not an option.
To be clear, generative AI poses real threats to creators and to the overall media and entertainment business (I frequently write about that sobering reality). But it’s also critical to assess transformational new technology stoically and learn to leverage it to expand possibilities, much like the entertainment industry ultimately did with Internet streaming. After all, generative AI’s box - like the Internet’s before it - is open and here to stay. We must all learn to live with it.
We are less than 15 months since OpenAI unleashed ChatGPT into a largely unsuspecting world. We correctly see the perils it poses to the creative community and the real human jobs that underpin it. But it’s also critical to accept, consider, and experiment with generative AI’s entirely new possibilities because that too is a form of creativity — so long as we never lose sight of the humanity at the center of it all.
What do you think? Send me your feedback and reach out to me at peter@creativemedia.biz and check out my firm Creative Media.
III. the cocktAIl - your curated AI mix (cool “must attend” AI events, helpful resources, alternative perspectives)
After all, it’s always happy hour somewhere!
Today’s curated cocktAIl - from your resident mix-AI-logist (yours truly!) - features three “must attend” generative AI-focused media and entertainment industry events.
“MUST ATTEND” events - save the dates!
(1) Digital Hollywood just announced its first exclusively generative AI-focused virtual summit, “The Digital Hollywood AI Summer Summit,” on four days in July (July 22nd - 25th). The depth and breadth of the sessions are outstanding, and they’re all virtual (so all of you can and should attend). Learn more here via this link.
Exciting related news. My firm Creative Media is joining forces with Digital Hollywood to host a series of free monthly generative AI-focused webinars leading up to the event to both prep you for (and preview) the Summit. I’ll interview leading artists and entrepreneurs about the state of generative AI (& their candid perspectives about it all). Look for more information about this new webinar series in upcoming newsletters.
(2) AI LA “Policy Happy Hour” March 14th, 2024, downtown LA. A good opportunity to mix with others in AI’s “in” crowd. It’s entirely free. Register here via this link. AI LA is one of the best AI-focused organizations in LA - deeply connected.
(3) AI LA’s “A.I. on the Lot” May 16th, 2024 event. This is also a “must attend” event. Check it out and sign up here. I attended last year. It’s where all the LA-based media and generative AI “movers and shakers” meet, learn, and collaborate. You can be one of them. I may even be able to get you a discount ….
Reach out to me at peter@creativemedia.biz with your feedback & submissions.
IV. the AI legal case tracker - updates on the key AI copyright infringement litigations
Here are the latest updates (as of 2/24/24).
(1) The New York Times v. Microsoft & OpenAI
Background: The Times filed its lawsuit in a New York federal court (Southern District) on December 27, 2023 for mass copyright infringement and other related claims), asserting that OpenAI “trained” its generative AI on “millions” of its copyrighted articles without consent and payment (and Microsoft enabled the infringement). In its pleading tour de force (one that serves as a master class for all future plaintiffs in these types of AI infringement cases), The Times also asserts that Microsoft and OpenAI are trying to build a “market substitute” for its news and, also notably, that their AI generates “hallucinations” based on The Times’ articles that substantially damage its reputation and brand. The Times seeks “billions of dollars of statutory and actual damages.” Microsoft and OpenAI will claim “fair use” - i.e., no license, payment or consent is needed.
Current Status: The Sarah Silverman plaintiffs in the OpenAI case discussed below (which is in a California federal court) are asking the judge here in this New York Times case to dismiss, pause or transfer it to California because similar issues are at play (and Silverman and the other artists believe the California court will be more sympathetic to their claims). Read my deep dive analysis of - and prediction for - the case and its ultimate resolution here.
(2) Sarah Silverman, et al. v. OpenAI
Background: Comedian Sarah Silverman and other artists filed this class action lawsuit in a California federal court (Northern District of California), asserting copyright infringement claims similar to those in The New York Times case (above) and in their separate California federal court case against Meta (below). As mentioned above, they are asking the judge in The Times’ New York case to put a halt to that case until this one plays out due to their similar issues.
Current Status: On February 12, 2024, the federal judge dismissed most of Silverman’s claims against OpenAI, rejecting her argument that the content generated by ChatGPT (i.e., the “output”) infringes her copyrighted works because there is no “substantial similarity” (and, therefore, no meaningful harm). But the court gave Silverman until March 13th to amend the complaint to plead a more direct link of harm to her and the other creators. Importantly, the court hasn’t yet ruled on the “input” side of the equation - whether OpenAI’s mass “training” of its generated AI on their copyrighted works without consent is infringement.
My prediction is that the judge here won’t change his mind and that he’ll permanently dismiss all claims on the “output” side of the equation. I also predict he will dismiss most, if not all, claims based on OpenAI’s unlicensed training on the “input” side. My advice to Silverman’s attorneys is to emulate The New York Times’ lawsuit against Microsoft and OpenAI. Read my deep dive analysis of the case, the court’s initial rulings, and what they portend here.
(3) Sarah Silverman, et al. v. Meta
Background: In this similar class action to her lawsuit filed against OpenAI, Sarah Silverman and others sued Mark Zuckerberg’s Meta on July 7, 2023 in another federal court (again Northern District of California) for mass infringement - i.e., unlicensed “training” of their generative AI on millions of copyrighted works, including their own. Not surprisingly, Meta’s defense is “fair use.”
Current Status: Much like in Silverman’s case against OpenAI, and for similar reasons, on November 2023, the judge dismissed the bulk of her claims against Meta. But much like in the OpenAI case, the judge gave Silverman a chance to amend her complaint to add a more direct link to actual harm (and Silverman filed her amended complaint in December). The re-hearing on Meta’s attempt to dismiss Silverman’s lawsuit is pending.
Much like in the OpenAI case above, my prediction is that the judge here won’t change his mind and most of Silverman’s case will be dismissed pre-trial. Read my deep dive analysis of the case, the court’s initial rulings, and what they portend here.
(4) Universal Music Group, et al. v. Anthropic
Background: UMG, Concord Music and several other major music companies sued Amazon-backed OpenAI competitor Anthropic on October 18, 2023 in a Tennessee federal court (Middle District of Tennessee). The music companies assert that Anthropic is infringing their music lyric copyrights on a massive scale by scraping the entire web to train its AI, essentially sucking up their copyrighted lyrics into its vortex – all without any licensing, consent or payment.
Current Status: Universal Music Group and the others have filed preliminary injunction motions (essentially asking the judge to stop Anthropic in its tracks from more non-consensual “training”), and Anthropic is asking the judge to dismiss their infringement claims. No hearings are yet scheduled on those motions, but the judge set the trial date for November 18, 2025 (no, that’s not a typo; 2025 it is). Read my deep dive analysis of - and prediction for - the case and its ultimate resolution here.
(5) Getty Images v. Stability AI and Midjourney
Background: Getty Images sued generative image AI companies Stability AI and Midjourney on February 3, 2023 in a Delaware federal court for mass infringement of its copyrighted photographic library. Getty’s claims are similar to those in The New York Times v. Microsoft & OpenAI case above, but here they are in the context of visual images instead of written articles - i.e., unlicensed scraping by their AI with an intent to compete directly with, and profit fro, Getty Images (i.e., market substitution).
Current Status: Unlike in the two Silverman cases above, the judge here so far has denied the defendants’ motion to dismiss the case. But the judge gave defendants a chance to try again after some procedural matters are resolved. Read more about the case and its background and overall context here.
NOTE: You’ll see the “AI legal case tracker” tab at the top of the page at “the brAIn” website. That’s where you’ll always be able to easily find and track the cases that matter.
Check out my firm, Creative Media’s, AI-focused services
Reach out to me at peter@creativemedia.biz to learn more and give feedback.