OpenAI IP promises ring hollow to Sora losers (www.theregister.com)

OpenAI’s Sora 2 — the free, ChatGPT-linked video generator that rocketed to the top of Apple’s App Store — has provoked a backlash from studios and talent after users began creating videos featuring copyrighted characters and real people without permission. In response, CEO Sam Altman announced tighter controls in a blog post: a “cameo” feature where people can upload video/audio of themselves to grant or limit use of their likeness, plus more granular settings for rightsholders to specify how characters can be generated. Altman also floated eventual revenue sharing for IP owners, but provided few specifics about permission workflows, enforcement, or whether banned material could still be used for model training. The announcement is significant because it highlights the unresolved tension between rapid generative-AI rollout and copyright accountability. Technical and policy gaps remain — black-box generation can produce unauthorized edge-case outputs, ownership chains complicate enforcement, and it’s unclear whether rights management will be opt-in, opt-out, or effective at scale. Major players like Disney and WME have already opted out, underscoring industry distrust. Comparing OpenAI’s posture to early YouTube — permissive release followed by retroactive monetization controls — the promised revenue share may be hollow unless OpenAI can reliably monetize video generation and build robust provenance and rights-audit mechanisms.