The Director Who Isn't Hiding His AI — And Why Every Filmmaker Should Be Paying Attention When the credits rolled on Young Washington, sharp-eyed viewers spotted something: roughly 100 shots in the film had been augmented with generative AI. Screenshots hit social media. A controversy started brewing. There was just one problem with the outrage: director Jon Erwin had never hidden any of it. He may be the most openly pro-transparency director working with AI in Hollywood today — and the box office didn't blink. Young Washington opened to $20.8 million on a $20 million budget, beat Supergirl that weekend, became Erwin's biggest opening ever, and already has a sequel on the way. Between his recent IndieWire interview and his 49-minute presentation at AI on the Lot, Erwin laid out one of the clearest real-world playbooks yet for how AI actually fits into professional filmmaking. Here's what stood out. "Synthesizing," not generating Erwin avoids the word "generating." At his studio, Innovative Dreams, AI doesn't invent shots from nothing — it augments material the production already owns and already shot in camera: real wardrobe, real locations, real actors. The clearest example: a sequence where George Washington crosses an icy river. Putting actors in frigid water wasn't safe, so the crew filmed the real river, filmed the actors in a 50-foot trench, and used AI to marry the two. Real inputs, provably owned — location agreements, actor consent, wardrobe — the same way you'd prove chain of title on a script. That's what keeps the work copyrightable and ethical. VFX in real time, on set — not months later in post This is the part Erwin calls Realtime Hybrid Filmmaking, and it's genuinely new. Think Avatar meets The Mandalorian: performance capture without mocap suits or face dots, combined with virtual production on a volume stage — with AI tools stacked on top of traditional software like Unreal Engine and Nuke. At his Manhattan Beach facility, prompting stations sit on wheels so AI artists work right beside the crew. Footage hits the edit suite about 30 seconds after "cut" via camera-to-cloud. An editor can raise a hand mid-shoot and ask for a shot, and the team performs and generates it on the spot. Erwin calls it nonlinear filmmaking: shooting, generating, and editing in the same day, not necessarily in that order.