OpenAI’s new Sora app has taken the world by storm, proving that if you give people a tool to conjure videos from their wildest imaginations, they will download it… a million times in less than five days. This breakneck success, even faster than ChatGPT’s own legendary launch, comes with a catch: the internet is now flooded with AI-generated videos of everything from your favorite Pokémon to deceased celebrities, sparking a copyright and ethical firestorm that OpenAI is scrambling to put out.
The Meteoric Rise of a “Slop” Generator
In the tech world, a successful launch is often measured by how quickly you can clog up people’s social feeds. By that metric, Sora is a superstar.
- A Million Downloads, Invite-Only: Hitting over a million downloads in under five days is a massive feat, especially considering the app was (and still is) invite-only and initially limited to users in the U.S. and Canada. This suggests a level of pent-up demand usually reserved for new iPhone releases or the last slice of pizza.
- Topping the Charts: The app quickly soared to the number one spot on the U.S. Apple App Store, outpacing tech giants like Google Gemini and even OpenAI’s own ChatGPT. It seems the allure of generating videos is, for now, stronger than the allure of generating text.
- Beyond Just Video: Sora is powered by the Sora 2 model, which is a significant leap forward. It doesn’t just create visuals; it can generate synchronized dialogue and sound effects, making the videos feel more complete and, consequently, more convincing. A key feature driving its viral nature is “cameos,” which allows users to digitally insert themselves or their friends into any AI-generated scene after a quick verification process. Imagine finally starring in your own superhero movie, or perhaps a video of you grilling and eating Pikachu—more on that nightmare fuel later.
The Inevitable Deluge of Digital Weirdness
With great power to generate content comes great irresponsibility from a portion of the user base. The “collective creativity of humanity” that Sora’s boss celebrated quickly manifested in some legally and ethically fraught ways.
- The Deceased Celebrity Problem: The app has led to a flood of AI-generated videos of deceased public figures, such as Michael Jackson and Tupac Shakur. This trend prompted a public plea from Zelda Williams, daughter of the late Robin Williams, asking people to stop sending her AI-generated videos of her father. In response to such concerns, an OpenAI spokesperson stated that while there are “strong free speech interests” in depicting historical figures, authorized persons can request that the likeness of “recently deceased” public figures not be used. They did not, however, specify what counts as “recent,” leaving a rather large gray area.
- The Copyright “Grilling”: More common, and perhaps more legally perilous, are the videos featuring copyrighted characters from films, TV, and video games. Social feeds were quickly filled with characters from SpongeBob SquarePants, South Park, and Pokémon. In a moment of peak irony, a viral deepfake featured OpenAI CEO Sam Altman alongside Pokémon characters, with the digital Altman quipping, “I hope Nintendo doesn’t sue us”. In another, he was shown grilling and eating the beloved mascot Pikachu. One can only imagine the legal department’s collective shudder.
OpenAI’s Course Correction: From “Opt-Out” to “Pay-Up”
Facing immediate and intense backlash from Hollywood studios and major talent agencies, OpenAI quickly pivoted its strategy. The company is now trying to move from a model that assumed permission to one that seeks permission and shares the potential profits.
The table below summarizes the key differences in OpenAI’s approach to copyright issues.
This shift is a pragmatic attempt to turn potential legal adversaries into business partners. As CEO Sam Altman framed it, the company is hearing from rightsholders who are excited about this new form of “interactive fan fiction” but want control over how their characters are used. The proposed revenue-sharing model is akin to how platforms like YouTube share income with creators, though the specific mechanics are still being figured out.
The Road Ahead: Simulated Worlds and Real-World Lawsuits
Despite the rapid commercial success, Sora’s journey is just beginning, and the road is paved with legal and ethical potholes.
- The Legal Gauntlet Remains: OpenAI’s new policy primarily addresses the output problem—what gets generated. It does not resolve the fundamental input problem: the ongoing legal battles over whether companies infringed copyright by scraping the internet to train their AI models in the first place. The potential cost of these battles is astronomical, as evidenced by AI firm Anthropic agreeing to pay up to $1.5 billion to settle a lawsuit with authors.
- A Mission to Simulate Reality: Beyond the social app, OpenAI’s ambitions for Sora are vast. The company positions it as a foundational step toward “general-purpose world simulators,” which it believes is a critical milestone on the path to achieving Artificial General Intelligence (AGI). In their view, teaching AI to understand and simulate the physical world is key to building systems that can solve real-world problems.
- An Uncertain Business Model: For now, the Sora app is free with “generous limits,” but Altman has openly admitted that the company will “have to somehow make money” from video generation, as users are creating much more content than expected. Future plans may include giving users the option to pay to generate extra videos.
So, as Sora continues its explosive growth, the big question is whether rights holders will embrace Altman’s vision of a new, interactive fan fiction paradise, or if OpenAI’s legal team is about to become as busy as the app’s video generation servers. One thing is for sure: the conversation around AI, creativity, and copyright will never be the same.