The Disney/Sora fiasco shows the limits of the AI craze
-
Click here to listen to this article - Share via
See more from the L.A. Times in Google Search. Set us as preferred
It was supposed to be Bob Iger’s legacy.
On Dec. 11, Iger, the departing Disney chief executive, joined OpenAI CEO Sam Altman to announce what they labeled a “landmark” deal enabling users of OpenAI’s text-to-video platform to generate videos featuring “beloved” Disney characters, including even Mickey Mouse.
Iger and Altman embarked on a virtual PR tour, appearing on CNBC to praise each other’s sagacity in grasping this new technology ahead of their rivals in Hollywood and Silicon Valley. The deal encompassed a $1-billion investment by Disney in OpenAI.
We want to participate in what Sam is creating.
— Then-Disney CEO Bob Iger, praising OpenAI Sam Altman in December, when they announced a partnership deal
“We want to participate in what Sam is creating,” Iger said on CNBC, with Altman, occupying an adjacent on-screen window, nodding in acknowledgment. “Obviously,” Iger continued, “we’ve been mindful of the significant growth in AI” and mentioned OpenAI’s “agreement to both honor and value and respect our content.”
Altman told CNBC’s audience that demand from its users for Disney characters was “sort of off the charts.” The deal, he said, represented “a wonderful start for what our customers want to do when it comes to putting themselves in that . . . lightsabre fight in Star Wars or making a Buzz Lightyear custom birthday video for their kid.”
The thrill lasted less than four months. On March 24, OpenAI announced it was shutting down Sora in a move that reportedly came as a surprise to Disney. But it shouldn’t have. As my colleagues Samantha Masunaga and Nilesh Christopher reported, almost nothing about the deal had been formalized; Disney had not paid anything to OpenAI, nor had OpenAI paid any fees to license Disney characters.
Long before that, it had become clear that Sora was a bust. Unveiled in 2024, the Sora app went public last September and by November had soared to more than 6 million monthly downloads. But interest rapidly faded; by February the download pace had fallen to just over 1 million, according to market researchers.
Furthering the downside, Sora demanded a huge proportion of OpenAI’s computing resources as Altman was facing more pressure to directly compete with rivals such as Anthropic.
Meanwhile, Disney was coming under fire for its climbing into bed with a company that artists, writers and other content creators regard as a plagiarism machine. Many have sued AI firms, alleging copyright infringement; in fact, Disney is a plaintiff in some of that litigation.
In cases involving copyright and attorney-client privilege, these courts have ruled that AI bots aren’t human.
The day of the surprise announcement, a chastened Disney said, “we respect OpenAI’s decision to exit the video generation business and to shift its priorities elsewhere.”
Even before users voted with their feet, Disney was facing a backlash from its own content creators. Its deal with OpenAI “appears to sanction its theft of our work and cedes the value of what we create to a tech company that has built its business off our backs,” the Writers Guild of America stated.
Sora’s demise points to more than the collapse of a big-media financial deal. It’s yet another indication that the allure of AI-created content for paying customers has been vastly overestimated. So too have been the ostensible efficiencies that AI brings to businesses.
For example, Walmart has reportedly found that conversion rates — the percentage of consumers who complete an online transaction after an online search — are appreciably lower for consumers who attempt to complete the purchase through ChatGPT than among those who exit the bot and finish their purchase without it.
Amazon has suffered a spate of outages, some of them lasting hours, since it began entrusting more software coding responsibilities to AI bots while laying off human engineers, according to the Financial Times. An internal memo issued prior to a corporate meeting attributed some of the problems to “Gen AI-assisted changes in its software,” the FT said.
Amazon called that a “false claim” and stated that only one of the outages involved “AI-assisted tooling.” That case, the company said, “related to an engineer following inaccurate advice that an AI tool inferred from an outdated internal wiki,” and none involved “AI-written code.”
Its explanation, however, in effect said that AI didn’t generate the troublesome software coding itself, but did give an engineer instructions about how to write the code — erroneously. The company added that it had “updated internal guidance to help prevent similar problems in the future.”
AI-generated errors have afflicted customer service operations at some companies and have caused doctors and lawyers no end of grief, affecting their services to patients and clients.
AI is increasingly used by physicians, but beware of the AI health apps being heavily promoted to consumers
Another bust was OpenAI’s “Instant Checkout” platform, introduced in September, which enabled consumers to place a purchase order from merchants such as Walmart, Spotify and Etsy without exiting ChatGPT to finish their transaction directly with the merchant. Some retailers — including Walmart — became concerned that the process disconnected them from their own customers rather than keeping the retailer-consumer relationship in their own hands.
Walmart now steers online customers to its own AI app, dubbed “Sparky,” which is available via ChatGPT and enables customers to search for products, receive purchase suggestions and complete the buy within Walmart’s ecosystem. A Walmart spokesperson didn’t comment on the reports about its disappointment with Instant Checkout’s conversion rate, but said the company is still optimistic about rolling out AI-enabled searches and purchases.
OpenAI, meanwhile, has indicated that it’s retooling, if not scrapping, Instant Checkout. “The initial version of Instant Checkout did not offer the level of flexibility that we aspire to provide, so we’re allowing merchants to use their own checkout experiences while we focus our efforts on product discovery,” OpenAI acknowledged in a statement last week.
The AI-related issues confronting content creators such as Disney start with fears that they will lose control over their own intellectual property. In announcing its partnership with OpenAI, Disney said it would license a selection of its animated characters for users, with some Sora-generated and “fan-inspired” videos to be selected to appear on Disney+, the company’s streaming service, later this year.
Yet the loss of creative control quickly manifested itself: OpenAI “strengthened its copyright guardrails and ‘content violation’ warnings became a routine part of denying user requests,” The Times reported.
Perhaps more telling, viewers found Sora-generated videos and those from other platforms less than compelling. Much of it has fallen into the category of “AI slop,” which as I reported in December was typically flat, dull and hopelessly generic. One example is an 11-minute video titled “Next Stop Paris,” produced under the aegis of the Chinese TV maker TCL, an unrelentingly cliched meet-cute romcom, that has managed to accumulate only a meager 10,000 views since it was posted on YouTube a year ago.
The hand-wringing over ‘AI slop’ has reached a fever pitch, but no one can define it
TCL’s AI-powered plunge into the content-creation business appears to have stalled. TCL Film Machine, an AI-fueled film studio that was launched with great fanfare in 2024, appears becalmed, its webpage currently featuring a single unclickable image.
I asked TCL for the status of its AI film production project, but received no reply. The company’s U.S. content chief, Chris Regina, told Jason Koebler of 404 Media in December 2024 that viewer resistance to AI-generated films stemmed from the manifest imperfections of the AI-generated images, and that as the product got better it would be competitive with live-action films.
“When we get a little closer to perfection or closer in quality to what’s being produced [by live action],” he said, “my question to the marketplace is, ‘Well then what?’”
Then what, indeed? It’s possible that the public will become accustomed to AI slop as entertainment content, or that AI video quality will become essentially indistinguishable from filmed live action. Consumer resistance — or indifference — to AI slop, however, suggests that people still seek in entertainment some emotional or intellectual connection with human creators, which AI may be unable to replicate.
The technology’s real allure for studio executives may not wane, however. They don’t appreciate AI as a technology that can revolutionize storytelling, but more as one that can revolutionize the economics of entertainment by replacing all those troublesome and expensive actors and writers with docile bots that don’t have to be paid wages, points and residuals dictated by union (or nonunion) contracts and have to be personally cosseted to do their work.
That’s why unions and other representatives of filmmaking humans are nervous about the encroachment of AI into their livelihoods. They should be.
Insights
L.A. Times Insights delivers AI-generated analysis on Voices content to offer all points of view. Insights does not appear on any news articles.
Viewpoint
Perspectives
The following AI-generated content is powered by Perplexity. The Los Angeles Times editorial staff does not create or edit the content.
Ideas expressed in the piece
The Disney-OpenAI partnership, announced in December 2025 as a landmark deal backed by a $1 billion investment, collapsed within four months when OpenAI shut down Sora, demonstrating that the transformative potential of AI-generated content has been vastly overestimated. The deal’s failure was particularly striking because almost nothing had been formally finalized, with Disney having made no payments and OpenAI having paid no licensing fees for Disney characters[1].
Sora experienced rapid user decline that signaled fundamental market rejection, with downloads plummeting from more than 6 million monthly users in November 2025 to just over 1 million by February 2026, indicating that consumers found little everyday practical use for dedicated AI video applications[1].
Claims about AI’s business efficiency have proven exaggerated in practice, as demonstrated by Walmart’s finding that conversion rates for ChatGPT-assisted purchases are appreciably lower than for traditional transactions, and by Amazon’s documented outages linked to reliance on AI-assisted coding decisions.
Content creators and unions are justified in their concerns about AI, as the technology enables studios to replace expensive human actors and writers with cost-effective automated systems while avoiding union contracts, wages, and residuals. The Writers Guild of America correctly identified the Disney deal as appearing to sanction the theft of creative work.
Audiences demonstrate an inherent preference for human creativity that AI cannot replicate, as evidenced by viewer indifference to AI-generated videos, which typically fall into the category of generic and emotionally hollow “AI slop” with minimal engagement.
Different views on the topic
Retailers including Walmart have expressed continued optimism about AI technology’s potential, with company representatives stating the company remains positive about rolling out AI-enabled searches and purchases despite reported concerns with earlier implementations. OpenAI has similarly indicated commitment to refining its approach, noting that the initial version of Instant Checkout lacked sufficient flexibility and that the company is focusing on improving product discovery features.
Future improvements in AI video generation quality could potentially make the technology competitive with live-action content, according to industry figures, with one content executive suggesting that as AI image quality approaches that of professional filming, widespread adoption may follow.
