On a recent Netflix earnings call, Ted Sarandos proclaimed the company “all in” on artificial intelligence. It sounded decisive—at least until you read the fine print. Per new guidelines circulated to its production partners, Netflix’s vision of the algorithmic future does not include populating prestige dramas with fabricated news broadcasts, dream-casting a second homicidal doll for a Squid Game sequel, or prompting A.I. to “create a character with Meryl Streep’s nose.” (Glad we got that cleared up.)

But a deeper look inside the streamer’s dos‑and‑don’ts manual provides an early snapshot of the entertainment industry’s cautious march toward what it now calls “ethical A.I.” Lawyers and executives have been noodling through these protocols for years, of course, ever since GPT‑3 first rattled studio lunch tables with fears of automated writers rooms and machine‑generated tentpoles. But this is Hollywood, where William Goldman’s axiom that nobody knows anything remains scripture. Hence the instinct to inch forward, clipboard in hand, careful not to spook either the A.I. tech lords or the unions.

A MESSAGE FROM OUR SPONSOR

WATCH GREAT SERIES ON MUBI

Now featuring new hand-picked series, from Joe Wright’s timely MUSSOLINI: SON OF THE CENTURY to Cooper Raiff’s HAL & HARPER, starring Mark Ruffalo.

Get 60 days free

Netflix is basically telling its vendors that, yes, generative A.I. is perfectly acceptable for brainstorming and mock-ups, but don’t prompt it with copyrighted works or public figures, and certainly don’t go feeding entire scripts into Sam Altman’s digital maw. However, there will inevitably come a moment when something goes sideways, and at that point, the only meaningful question will be who built a compliance system sturdy enough to say they saw it coming and behaved like adults.

Put another way, A.I.’s age of innocence, when the phrase “we didn’t know” could at least buy you some time in Hollywood, has come and very rapidly gone. Consider Disney’s nearly decade‑long legal struggle over the Mova facial capture system. After visual effects vendor Digital Domain 3.0 allegedly used the facial capture tech on major Disney projects, Rearden, a Silicon Valley firm, claimed Digital Domain 3.0 had stolen it from them. Disney’s exposure ballooned into a string of lawsuits, including one tied to Beauty and the Beast that actually went to trial. Six weeks ago, the Ninth Circuit reversed a district court’s decision wiping a jury verdict, and held that Disney could be on the hook because it had the practical ability to supervise its vendor—in other words, the company could have performed due diligence on the core software. Two weeks ago, as I reported, a judge applied the same logic in a case tied to the Avengers films.

As the big studios partner with companies pushing the tech envelope in ways that often end up in court, it’s only a matter of time before someone starts poking around and asking exactly how these tools are being supervised. Better to get in front of the fire than pretend you can outrun it.

Here’s the funny thing, though, about vicarious liability: Hollywood wants, and increasingly needs, a more expansive view of responsibility for those standing close enough to the wrongdoing in order to nudge it, shape it, or stop it.

Take the Disney and Universal lawsuit against Midjourney, the A.I. image-generation company. The studios argue that even if the A.I. platform can’t be nailed for direct infringement over images its users generate, it should still be liable for failing to prevent the misconduct. After all, they note, Midjourney has shown it can exercise control—at various points, the company has blocked users from conjuring certain protected characters. The fact that those guardrails later loosened, the studios contend, reflects a business decision to continue profiting from infringement.

Midjourney maintains that most of what its users generate is non-infringing, and even when familiar faces like Darth Vader or the Minions appear, the purpose is often lawful: parody, commentary, fair use. The burden, they argue, should fall on rights-holders to identify specific violations before any action is taken. Besides, Midjourney’s attorneys add, Disney and Universal may “demagogue,” but they shouldn’t be allowed to “have it both ways” as many of their partners in the visual effects industry “have been utilizing Midjourney professionally and for [these studios’] benefit.”

The next chapter in this fight figures to be written at the Supreme Court. Oral arguments are set for December 1 in Cox v. Sony Music, following the headline-making, billion‑dollar verdict against Cox in 2019 that the Fourth Circuit later partially unwound. The justices will decide whether internet service providers must yank the accounts of users repeatedly flagged for piracy. Cox contends that unless it actively induced piracy, it shouldn’t be held responsible for the bad acts of a few subscribers.

The content industry, unsurprisingly, disagrees. In an amicus brief filed on October 22, the Motion Picture Association warned that Cox’s position is “destabilizing,” arguing that chasing down every individual infringer is impossible, and that the only way to mitigate harm is to incentivize internet companies to cooperate in the policing. In its own amicus brief, the Authors Guild goes further, stressing that unlike the VCR makers of yore, modern internet companies have ongoing relationships with users and knowingly continue to serve those who are “substantially certain” to infringe. This, they contend, crosses the line.

A MESSAGE FROM OUR SPONSOR

WATCH GREAT CINEMA ON MUBI

The studio & streaming platform behind Oscar-winning THE SUBSTANCE. Discover new hand-picked films, from Gotham Awards-nominated LURKER to THE HISTORY OF SOUND, starring Paul Mescal and Josh O’Connor.

Get 60 days free

Most A.I. players—OpenAI, Anthropic, Meta—are conspicuously sitting this one out. But Elon Musk’s X Corp. has jumped in, telling the justices that the federal judge overseeing New York Times v. Microsoft/OpenAI erred earlier this year when he let contributory infringement claims survive. A.I. has a multitude of lawful uses, argue X’s attorneys, and liability shouldn’t attach simply because a platform knows its technology might be misused. Without “proper guidance,” Musk & Co. warn, billion‑dollar verdicts could “wreak havoc” on the nascent A.I. ecosystem.

Meanwhile, Universal Music Group is attempting one of the more audacious pivots of the generative A.I. era: bowing out of its lawsuit against Udio, the buzzy A.I. song-generator startup, and signing on as a strategic partner instead. UMG had previously pursued the case against Udio in lockstep with Sony and Warner, and it was shaping up to be a barn burner, with the record giants accusing Udio of mass-scale copying of their sound recordings, while Udio, flanked by an all-star defense lineup (Alex Spiro, Sy Damle, et al.), fired back with the accusation that the majors were using their I.P. to fortify their decades-old monopoly on the business of making and selling music.

Rather than stick around for the possibility of a multibillion-dollar judgment—or the full airing of that copyright misuse counterpunch—UMG decided to zag. The company inked a deal with Udio for a new “walled garden” subscription platform, giving Udio a license to train on its catalogue. Days later, UMG chairman and C.E.O. Lucian Grainge was on the company’s earnings call, touting a sometime-in-2026 launch of an “ethically trained” A.I. music platform that would, he promised, “provide further revenue opportunities for artists and songwriters and UMG.”

That’s a nice story for investors, especially with UMG still eyeing a U.S. public listing and hunting for A.I.-adjacent revenue streams. But what is UMG getting in return? More to the point, how much of that is being shared with the artists?

Irving Azoff, Grainge’s peer in the music industry pantheon, told me he wants answers. Will UMG be receiving equity, he asks, like it did with Spotify back in the day? Will there be new income streams? If so, how will that money be split? “There’s no excuse for there not to be transparency if they’re using artists’ catalogues,” Azoff said.

For now, Universal Music Group is signaling that participation will be optional. Recording artists like Taylor Swift, Ariana Grande, and Sabrina Carpenter will reportedly have the choice to opt in to having their voice and likeness used to generate synthetic A.I. music—ethically, of course. But the fine print hasn’t been made public, and that matters because this deal will be a benchmark. As Sony and Warner push forward with their own lawsuits, this is the model they’ll be measured against. UMG declined my request for details.

Thanks, Eriq. See you on Thursday.

Matt

Need help? Review our FAQ page or contact us for assistance. For brand partnerships, email ads@puck.news.

You received this email because you signed up to receive emails from Puck, or as part of your Puck account associated with {{customer.email}}. To stop receiving this newsletter and/or manage all your email preferences, click here.

Puck is published by Heat Media LLC. 107 Greenwich St., New York, NY 10006