No one knew what was going on. It was late September, just days before OpenAI rolled out a new social media app for its video generator, Sora, and whispers were swirling of changes that would undercut Hollywood. A major talent agency, which learned of the upcoming news through the grapevine, wondered why its C-suite hadn’t heard from the Sam Altman-led start-up yet, according to an agency exec involved with the talks. The talent firm reached out.
The first round of talks between the two sides was contentious. OpenAI was “purposely misleading,” says this agency exec who was part of the heated back-and-forth. The tech giant personnel leading the discussions took an upbeat tone, repeatedly talking up an opt-in regime that would protect the agency’s clients against the misuse of their intellectual property and likenesses.
Hollywood will be happy with the improved guardrails on Sora 2.0, stressed OpenAI’s team, which included COO Brad Lightcap, Sora product lead Rohan Sahai, media partnerships vp Varun Shetty and talent partnerships lead Anna McKean, according to the exec. It turned out that wasn’t the case. “We started exchanging notes with others having similar conversations and realized we’re all hearing different things,” this person says.
Among the discrepancies: the treatment of likenesses versus intellectual property. Tellingly, some execs were told an opt-in would be required for both. Others were told the opposite, or weren’t notified of the distinction. OpenAI’s messaging was haphazard to Hollywood.
Around that time, WME was told in no uncertain terms that its clients, which include Matthew McConaughey, Michael B. Jordan and Ryan Reynolds, would have to explicitly notify OpenAI not to have their faces and voices appear on the platform, according to a WME partner involved in the discussions. He recoiled at the thought. “I said, ‘Imagine an agent calling a client right now and advocating for them to get onto Sora,’ ” the partner recounts. “It’s very likely that client would fire their agent. None of us would make that call.” In talks with OpenAI, the partner stressed to the tech firm, “We have a huge client roster. If they all opt out, what are you doing this for?”
On Sept. 29, the day before Sora 2 was released, this WME partner was told that likenesses wouldn’t be used without permission. It was a meaningful — albeit halfway — win for the agencies when OpenAI unveiled its new social media app, which allows users to create content featuring cash-cow properties owned by studios across Hollywood. Studios must individually flag the properties they don’t want to be used, an aggressive escalation of Silicon Valley’s encroachment onto the entertainment industry.
On Oct. 3, Altman announced that the company was pivoting to a kind-of-but-not-really opt-in regime. But by that point, the gambit had worked. At one point, the video generator was returning clips that feature recognizable movies, TV shows and games, including Bob’s Burgers, SpongeBob SquarePants, Gravity Falls, Pokémon, Grand Theft Auto and Red Dead Redemption, among many others. Sora is now the most popular free option on the App Store.
“This was a very calculated set of moves he made,” the agency exec says. “They knew exactly what they were doing when they released this without protections and guardrails.” Talks, now involving legal personnel, are underway. Litigation is being considered.
As OpenAI races ahead, Hollywood is scrambling to get in front of the situation. The typically tight-lipped Motion Picture Association, the industry’s top lobbying group that mostly has stayed on the sidelines as the AI fight plays out, huddled up and saw it as necessary to publicly rebuke the company. So did CAA and UTA, with WME head of digital strategy Chris Jacquemin circulating a memo to agents saying that he had alerted the tech giant that none of its clients were participating in the latest version of Sora.
Internally at studios and agencies, there’s confusion — and frustration — around the opt-out mechanism, according to people familiar with the matter, all of whom noted there’s no formal system for doing so and characterized Altman’s walk back as lip service. Individual properties are flagged and taken down one by one by way of a link that acts more like a mechanism for reporting infractions, a slow and cumbersome process, the agency exec says, noting that OpenAI had no dedicated personnel responsible for guardrails on Sora leading up to its release, though it has a few now. Among some Hollywood execs, there’s a new reluctance to even engage with OpenAI.
They’re of the belief that entering into those talks will now and forever put the onus on them to notify the company that their intellectual property can’t be used. In a sharply worded letter sent to OpenAI in late September, Disney said it “is not required to ‘opt out’ of inclusion of its works” to “preserve or pursue its rights under copyright law.”
Earlier this year, the studio, along with Universal and Warner Bros. Discovery, launched the first of what’s increasingly likely to be a series of lawsuits over AI firms’ indiscriminate use of its intellectual property when it sued Midjourney for allowing users to create content featuring copyrighted characters.
“They’re turning copyright on its head,” says Rob Rosenberg, partner at legal advisory firm Moses Singer and former Showtime Networks executive vp. “They’re setting up this false bargain where they can do this unless you opt out. And if you didn’t, it’s your fault.”
(Left to Right): OpenAI’s COO Brad Lightcap and CEO Sam Altman at the Sun Valley Allen & Co retreat in July.
Getty Images
For years, OpenAI’s cavalier approach to intellectual property rights in Hollywood, led in part by tech industry veterans Shetty and McKean, has been consistent: Ask forgiveness, not permission. It’s the path of least resistance to monetize Sora. Use cases for ChatGPT are far and wide, with the company generating roughly $1 billion a month off the back of roughly 700 million weekly active users for its chatbot products, reported The Information.
It’s murkier for Sora. AI video generators’ existence depends in large part on their ability to spit out recognizable properties. Consider Midjourney instituting safeguards to prevent the generation of copyrighted material after it was sued but removing them shortly after engagement dropped. Altman is putting a megaphone on his plans to exploit Sora, and it turns out it’ll depend on whether studios are willing to engage in licensing.
“We are hearing from a lot of rightsholders who are very excited for this new kind of ‘interactive fan fiction’ and think this new kind of engagement will accrue a lot of value to them,” Altman wrote in a blog post.
Shetty echoed the sentiment, “We see new opportunities for creators to deepen their connection with the fans.”
Hollywood is fighting a battle of attrition with a well-capitalized AI industry, and it’s losing ground, much like what happened at the dawn of the internet when it was too slow to combat piracy. The biggest piece of leverage it has is the prospect of future deals. OpenAI is entering into those conversations as more of an adversary than an ally.
“How are you coming to the industry expecting partnership?” the WME partner recounts telling OpenAI personnel. “You quite literally set the bridge on fire.”
Inside the agencies, there’s some resentment over studios’ reluctance to challenge Silicon Valley’s steady intrusion onto Hollywood, according to people familiar with the matter. AI companies have been training their technology for years on movies and TV shows scraped across the internet. There’s little recourse for creators, most of whom don’t actually own the intellectual property in their underlying works since they’re mostly commissioned on a work made for hire basis.
What if the studios had turned to court earlier? Could they have secured a deal that forced AI companies to delete stolen content from their training libraries, much like the settlement authors secured from Amazon-owned Anthropic? Right now, studios are hyperfixated on defending their intellectual property rather than further exploiting it with AI partnerships. It’s a balancing act. Some may have future plans to deploy the technology in ways that could be undercut depending on the legal battles they pick.
Disney could decide to forego such AI deals and create a standalone app, much like OpenAI’s Sora, in which users pay a monthly subscription fee to generate interactive content of its properties. Others may have in dealmaking in their sights. It’s here the studios’ interests may not align with the agencies.
“Therein lies the big conundrum,” the WME partner says. “If they sue, they cut themselves off from being able to partner with these companies.”
The rival agency exec was less forgiving: “Our clients expect to be protected. There are issues at hand that this is bringing to light.”
Sora 2 can easily create short, hyperrealistic clips in a variety of styles.
This story appeared in the Oct. 15 issue of The Hollywood Reporter magazine. Click here to subscribe.
First Appeared on
Source link
