Two Contradictory Things Are Both True

In 2026, Hollywood is simultaneously suing AI companies over unauthorised use of copyrighted material and quietly building its own AI capabilities as fast as it can. Reporting from the Los Angeles Times found that among hundreds of studio job postings in a single month, more than one in ten was connected to AI - split between teams building internal AI tools and teams built specifically to defend the studio's intellectual property against unauthorised AI use elsewhere. As one industry insider running an AI tooling company for studios put it, plenty of studios are hiring for AI capability but simply choosing not to discuss it publicly, wary of consumer and union backlash.

The Fights That Made Headlines

OpenAI's Sora video generator triggered the first major flashpoint in late 2025 when users generated convincing clips using Hollywood characters and celebrity likenesses without consent, prompting the Motion Picture Association to publicly demand "immediate and decisive action." A second, larger wave hit in early 2026 when TikTok owner ByteDance released Seedance 2.0, which users immediately turned toward generating unauthorised content featuring named actors - including a widely circulated fabricated fight scene between two real, still-living movie stars - reigniting the same alarm across talent agencies and unions. In both cases, the affected companies eventually offered more granular opt-out and consent controls for rights holders, but only after the content had already spread and the backlash had already landed.

What the Unions Actually Won

Hollywood's major unions didn't just protest - they negotiated real, enforceable protections during recent labour contract cycles. The Writers Guild secured provisions keeping AI out of the writers' room process itself. SAG-AFTRA won limits on how much work fully synthetic, AI-generated performers can do in place of human actors. The Directors Guild secured a requirement that production companies must inform directors in advance if they'll be required to use AI tools on a given production - giving directors visibility and some leverage that didn't previously exist. These aren't symbolic wins; they're specific, contractually binding constraints that reshape how AI can legally be used in unionised productions.

The High-Profile Deal That Fell Apart

Disney's widely publicised, roughly $1 billion multi-year partnership with OpenAI - including Sora licensing rights for over 200 Disney characters - collapsed entirely within months of being announced. "The deal is not moving forward," then-CEO Bob Iger later confirmed. It's a genuinely instructive case study: even one of the industry's most resourced, AI-forward studios pulled back from its most ambitious public AI bet, and there's still no clear evidence, even among companies that have kept experimenting, that generative AI is measurably saving Hollywood money or improving output at the scale the initial hype suggested.

Where AI Is Quietly Actually Working

Away from headline-grabbing generative video tools, AI has found genuine, less controversial footholds in production: visual effects work, storyboarding, sound design, dialogue editing, and de-aging or voice restoration for productions needing to work around an actor's unavailability or, in some documented cases, death. These uses attract far less public attention than "AI writes your movie," but represent the more durable near-term reality - AI as a production tool embedded in specific technical workflows, not as a replacement for the creative direction of a film.

The Honest Read on Where This Goes

The evidence so far doesn't support either the most alarmist prediction ("Hollywood is over") or the most optimistic one ("AI will revolutionise filmmaking"). What's happening instead is a slower, messier, more contested integration - real technical adoption in specific workflows, genuine and successful union pushback constraining the more disruptive uses, high-profile partnership announcements that frequently don't survive contact with reality, and a persistent, unresolved legal fight over consent and compensation for using real people's likeness and copyrighted material in AI training and output. That fight, not any single technology breakthrough, is likely to be the thing that actually determines how AI reshapes the industry over the next several years.