OpenAI and Hollywood studios clash over copyrights and consent (www.latimes.com)

OpenAI’s Sora 2—an upgraded video-generation tool that lets users upload footage of real people and place likenesses (including voiced dialogue and sound effects) into AI-created scenes—has provoked an immediate backlash from Hollywood studios, talent agencies and unions. Demonstrations showing a synthetic Michael Jackson taking a selfie alongside a Bryan Cranston lookalike and a SpongeBob likeness in the Oval Office crystallized industry fears. Trade groups (MPA), major agencies (WME, CAA, UTA) and SAG‑AFTRA argue OpenAI’s approach—reportedly treating many performers as included unless they opt out—violates copyright and publicity rights and threatens compensation frameworks; OpenAI says it’s adding granular controls, takedown review processes and working on ways to compensate rights holders. The clash matters because it intersects core technical capabilities (video-conditioned generative models that can mimic real faces, voices and characters) with longstanding IP law and commercial models for talent and franchises. If studios prevail, platforms will likely need opt-in licensing, attribution, or new revenue-sharing systems; if tech firms keep a permissive opt-out posture, expect widespread litigation and aggressive takedown enforcement. Practically, this will shape dataset curation, model guardrails, consent mechanisms, and the economics of creative AI—setting precedents for how likeness, copyrighted characters and compensation are handled across the entertainment ecosystem.