Hollywood backlash grows against OpenAI's new Sora video model

Key Concepts Sora video model, Copyright Law, Intellectual Property (IP), Fair Use, Generative AI Outputs, Training Data, Guardrails, Back Door Prompts, Granular Controls, Revenue Sharing.

Hollywood's Backlash Against OpenAI's Sora

Hollywood studios are expressing significant outrage following OpenAI's introduction of its new Sora video model. The initial controversy stemmed from OpenAI's statement that Sora would utilize entertainment companies' copyrighted material unless they explicitly opted out. This approach was met with strong condemnation from movie studio executives, who deemed it a fundamental misunderstanding of how copyright law operates. Sources indicate that executives were "outraged that OpenAI thought it could ask them to opt out of their intellectual properties inclusion in Sora. Because that's just not how copyright law works."

Following this backlash, OpenAI CEO Sam Altman "shifted tone," promising to provide "much more granular controls" to rights holders. However, studios remain vigilant, actively assessing the extent to which Sora could infringe upon their copyrights.

Demonstrations of Copyright Infringement

Studios are actively investigating potential copyright violations. A source revealed that Warner Brothers engaged with OpenAI, presenting evidence of both "blatant copyright infringement" and instances of infringement via "back door prompts" that do not explicitly name protected characters.

• The Batman-like Character Example: A specific test involved prompting Sora to "create a superhero that wears a black cape and is saving a woman from a burning building." The resulting video quickly generated a character that "looks and sounds very similar to Batman," demonstrating Sora's ability to produce content closely resembling copyrighted characters without direct instruction.
• Garfield-like Character and Sora's Guardrails: In contrast, attempts to create a "Garfield like character" or a "fat orange cat eating lasagna" were blocked by Sora's "guardrails." Instead, the tool produced a video of a "tabby cat licking a lasagna." This suggests that some modifications and protective measures have been implemented in the tool since its launch, though studios are closely monitoring OpenAI's ongoing response to their concerns.

Studios' Strategic Responses and Concerns

Studios are currently prioritizing negotiation over immediate litigation, preferring to resolve issues through dialogue. However, they are prepared for legal action if necessary.

• Revenue Sharing Offer Rejection: Sam Altman's offer to share revenue with studios whose IP is used by Sora has largely failed to appeal to many media companies. Their primary concern is not just financial compensation but maintaining "control how their characters show up" and ensuring the integrity of their intellectual property.
• Focus on Training Data vs. Generative Outputs: An entertainment attorney highlighted a critical concern: the current focus on "generative AI outputs" might be "distracting from the question of whether Saw used copyrighted materials for training data." This perspective suggests that the core issue might lie in the source material used to train the AI model, rather than solely the output it produces. OpenAI maintains that its policies regarding Sora adhere to "fair use" principles, but the complexity of this issue is widely acknowledged.

Legal Precedents and Future Implications

The potential for further lawsuits is significant, especially given past legal actions. Universal Pictures, Disney, and Warner Brothers have previously sued other generative AI companies like Midjourney and Minimax, establishing a precedent for studios to pursue legal recourse against perceived copyright infringement by AI models.

Conclusion: A Complex and Evolving Landscape

The situation between Hollywood studios and OpenAI's Sora model is a complex and evolving challenge at the intersection of technology, creativity, and intellectual property law. While OpenAI has shifted its stance towards offering more granular controls and revenue sharing, studios remain deeply concerned about copyright infringement, the control over their characters, and the fundamental question of how AI models are trained using copyrighted material. The industry is closely watching how OpenAI addresses these concerns, with a clear preference for negotiation but a readiness to litigate, underscoring the significant legal and ethical implications of generative AI for the entertainment industry.