TLDR: OpenAI’s recently launched Sora 2 video generation model is facing significant backlash and raising serious copyright concerns. The AI tool is capable of creating high-definition videos featuring popular copyrighted characters like Pikachu, Goku, and SpongeBob SquarePants, often in unusual scenarios. Critics argue that the model’s training on copyrighted material without explicit consent, and its ‘opt-out’ policy for rights holders, fundamentally misunderstands copyright law, leading to potential legal challenges and a wider debate on intellectual property in generative AI.
OpenAI’s latest advancement in generative artificial intelligence, Sora 2, has quickly become a focal point of both fascination and controversy since its launch on September 30, 2025. The sophisticated text-to-video model is capable of generating high-definition, short-form animated clips up to a minute long, showcasing a remarkable ability to interpret prompts and create compelling visual narratives. However, this impressive capability has inadvertently plunged OpenAI into a heated debate over copyright infringement and intellectual property rights.
Reports indicate that users of Sora 2 have been generating videos featuring a wide array of copyrighted characters from popular franchises. Examples include Pikachu stealing Poké Balls, Goku shaking hands with Mario, SpongeBob SquarePants in various inappropriate scenarios (including a ‘Nazi SpongeBob’), Lara Croft fighting Kratos, Master Chief pursued by police, and even pixel-perfect scenes from ‘The Simpsons’ that do not officially exist. These creations, often depicting characters in unusual or unauthorized contexts, have triggered alarm bells among copyright holders and legal experts.
A core issue lies in Sora 2’s training methodology. The system is reportedly trained on vast amounts of copyrighted material by default, placing the onus on rights holders to request the removal of infringing content rather than requiring their consent in advance. This ‘opt-out’ approach is being heavily criticized. As one expert noted, ‘If OpenAI is taking an aggressive approach that says they’re going to allow outputs of your copyright-protected material unless you opt out, that strikes me as not likely to work. That’s not how copyright law works. You don’t have to opt out of somebody else’s rules.’ This stance positions OpenAI at the center of a broader legal battle concerning generative AI and intellectual property, with similar lawsuits already underway against other AI companies like Midjourney by entities such as Disney, NBC, and Warner Bros.
Beyond copyrighted characters, Sora 2 has also demonstrated the ability to generate ‘deepfakes’ of public figures and ordinary users, despite OpenAI’s claims of implementing safeguards. Some users have reported seeing manipulated content involving real celebrities, including a controversial instance of a blackface version of Scarlett Johansson. This echoes a previous controversy where OpenAI faced criticism for using a voice resembling Johansson in its ChatGPT product.
Also Read:
- Perplexity AI Acquires Visual Electric to Bolster Generative AI Capabilities Against Industry Giants
- OpenAI and Japan’s Digital Agency Forge Alliance for AI Integration in Public Services
Critics are also voicing concerns about the potential impact on human animators, suggesting that tools like Sora 2 might encourage studios to reduce their reliance on human talent in favor of cheaper AI-generated outputs. The rapid advancement of AI video generation, exemplified by Sora 2, is being likened to a ‘Pandora’s Box moment’ similar to the initial release of ChatGPT, blurring the lines between reality and computer-generated content. While OpenAI states it is developing detection classifiers and plans to include C2PA metadata to indicate AI-generated videos, the ease with which such metadata can be removed raises further questions about the long-term implications of this technology.
