Sam Altman is squaring off against Hollywood with the new Sora (www.businessinsider.com)

OpenAI has told Hollywood studios and talent agencies that its Sora video-generation system can produce material based on existing characters and copyrighted works without needing explicit permission, and that rights holders who object must opt out case-by-case. The Wall Street Journal reports OpenAI is effectively proposing a default “use unless you object” regime for copyrighted content, a sharp contrast to prior platform norms where tech companies accepted user uploads but relied on takedown systems and licensing deals to manage copyright risk. If this posture holds, it would be a major legal and business inflection point for AI/ML: models trained on broad swaths of copyrighted media would not only learn from those works but could be used to synthesize new, derivative scenes or characters unless proactively blocked. That raises thorny questions about whether generative outputs that recreate or closely mimic protected material violate copyright, and shifts the burden to content owners to police use. Technically, it underscores how multimodal generative models (text-to-video/imagery) change the calculus of dataset use, licensing, and content-moderation systems. Expect courtroom tests, new licensing frameworks, and shifts in how studios negotiate access — similar to how YouTube’s early “ask forgiveness later” growth forced later industry-wide settlements.