When Sora was released many assumed it was “the death of Hollywood.” Only two years, then Sora no longer exists

In February 2024, OpenAI published on X a string of AI-generated videos with his new model, Sora. Although today, after two years of progress, they even feel outdated, at the time the result was convincing enough for the media around the world to start headlines that Hollywood had a very serious problem. Two years later, Sora does not exist.

Panic effect. The effect of this presentation with videos was immediate: MIT Technology Review, for example, described them as “impressive“, although warning that they had probably been chosen and were not representative of the output usual. That did not stop the narrative: for weeks, the dominant conversation in the specialized media was that film studios were facing an almost perfect replacement tool: synthetic actors, sets generated in seconds, automated post-production… The Hollywood unions, which they had signed agreements with the studios the previous year after a historic strike they put the issue back on the table.

Two bombs. Sora’s story has two moments of media panic, separated by eighteen months. The first arrived in February 2024, with the presentation of the model described above. There was talk that Hollywood had a serious problem, that the almost perfect replacement tool already existed and that the studios were not prepared to face this threat. The second came with the launch of Sora 2 in September 2025with real faces inserted in videos generated by AI and with third-party intellectual property by default, unless the prompts expressly requested otherwise. All of this multiplied the volume and intensity of the alarm in Hollywood and the media.

What was said In February 2024, coverage of Sora’s first model mixed amazement and alarm in similar proportions. Fortune commented that OpenAI had moved the generative AI battle directly to Hollywood. NBCNews asked filmmakers if this was the end of Hollywood, and some responded that it wasn’t yet. IndieWire He sensed that Sora could mean the apocalypse of cinema.

The cycle of apocalyptic headlines with Sora 2 was much more intense. CNBC declared that the app was challenging Hollywood and causing panic in the film industry. deadline He said Hollywood was raw. LA Times He spoke of a battle that was worsening and a firestorm unleashed in the sector. slatewell, he talked about how AI was about to crush Hollywood as we had known it.

What happened then. The panic increased in December 2025, when Disney, the most careful entertainment company in the world with its intellectual property, signed a three-year agreement with OpenAI: investment of 1 billion dollars and access to more than 200 characters from Disney, Marvel, Pixar and ‘Star Wars’ so that Sora users could generate them in their videos. Disney+ would broadcast a curated selection of that content. It was the definitive legitimation, which has only lasted 90 days. OpenAI has closed Sora before a single dollar has changed hands.

Property problems. Sora’s problems have not only been financial. The app has accumulated a long list of controversies: deepfakes of deceased public figuresmassive use of copyrighted characters without permission prior, and the appearance of external tools to remove watermarks that identified AI-generated content. In November 2025, CODA (Japanese association representing, among others, Studio Ghibli and Square Enix) sent a formal letter to OpenAI demanding that it stop using its intellectual property to train the model. The families of Robin Williams and George Carlin They publicly asked for it to be blocked generating videos with your images. Moderating generative video content at scale turned out to be much more complex than moderating text or image.

The consequences of hype. Analyst Ed Zitron criticized this attitude of the media, stating that they did not cover the launch of Sora but rather they amplified their marketing. Saying that Sora was a real threat to Hollywood was, from the beginning, an extrapolation built on selected demos and clips of a few seconds. Thousands of audiovisual professionals spent months convinced that their industry was about to be replaced by a tool that, according to OpenAI’s own numbers, never found enough users willing to pay $200 a month for it.

The hype cycle has real consequences: it inflates expectations that are not met, generates costly defensive decisions, and when the product closes, no one takes critical stock. Sora’s coverage is a textbook case of how uncritical amplification of tech demos can be confused with industry analysis, and the damage that attitude can do.

Hollywood is still alive. The closure of Sora does not erase the generative video sector in one fell swoop: runwaywhich rejected an acquisition offer from Meta, currently leads the sector with its Gen-4.5 model, along with I see 3.1 from Google and Chinese models Kling and Seedance. These tools are absorbing the space that OpenAI abandons. Who no one absorbs is Hollywood. The film industry, with all your problems (reorganizations, box office decline, threat of streaming), remains a profitable business built on decades of well-established creation, distribution chains and franchises that no generative model can replicate with a prompt.

The question is not whether AI will transform audiovisual production (it is already doing so, in post-production, pre-visualization and marketing content creation) but in what real time frames and under what viable economic models. For now, the market responds that generating photorealistic video on a massive scale is computationally very expensive and that consumer users are not willing to pay what it costs. Disney signing Sora wasn’t evidence that Hollywood was in danger. It was, rather, evidence that big studios want to be in the AI ​​conversation, not outside of it.

In Xataka | Seedance’s strategy was to copy first, go viral later and back away later. Until Hollywood said “enough”