Ground Truth.
AI, checked against the source.

News · 2026-06-22

Google DeepMind puts $75 million into film studio A24 to build AI moviemaking tools

AI's collision with Hollywood usually shows up as a fight -- over jobs, over likeness rights, over whether a model was trained on someone's work without permission. This week it showed up as a partnership instead. Google DeepMind, one of the world's leading AI labs, is investing around seventy-five million dollars in A24, the independent studio behind a string of acclaimed, distinctive films, to jointly develop tools for making movies (Deadline; Reuters and other major outlets corroborate the deal). The plan, as described, is for DeepMind's researchers to work directly alongside filmmakers, building and refining production tools in the actual messy context of making a movie rather than in a lab.

The most important detail is what the companies say it is not. Officially, this is not a deal for Google to train its AI models on A24's catalog of films -- not a data-licensing arrangement dressed up as a collaboration. It's framed as a tooling and workflow partnership: figuring out where AI can genuinely help in the craft of filmmaking, from pre-visualization to editing to the countless tedious steps in between, by embedding researchers with the people who actually do the work.

Here's the background that makes this interesting. AI labs are very good at building general-purpose tools and often quite bad at knowing what professionals in a specific craft actually need. A filmmaker doesn't want 'generate a video from a prompt' as much as they want help with the specific, unglamorous problems of their day -- matching shots, planning scenes, handling the thousand small decisions a production runs on. The only reliable way to learn those needs is to be in the room. By buying a stake in a respected studio and putting researchers on real productions, DeepMind is trying to shortcut the gap between 'powerful AI' and 'AI that filmmakers actually want to use.'

Think of it as the difference between an engineer designing kitchen equipment from a spec sheet versus one who spends six months working the line in a busy restaurant. The second engineer builds better equipment because they've felt the actual problems. DeepMind is, in effect, buying its way onto the line.

It's worth remembering how charged the backdrop is. The relationship between AI and the film industry has, until now, mostly been adversarial -- a major driver of recent labor disputes was fear that studios would use AI to replace writers, actors, and crews, or to train models on people's work and likenesses without consent. A frontier lab investing in a studio to build tools with filmmakers is a deliberate attempt to write a different story: AI as a collaborator that handles the tedious, expensive parts of production rather than a replacement for the people who do the creative work. Whether it actually lands that way depends entirely on how the tools are built and who benefits -- which is exactly why the details matter more than the press release.

Why it matters: this is a sign of how the next phase of AI competition plays out -- not just who has the best model, but who has the deepest hooks into specific high-value industries. Owning a relationship with a prestige studio gives Google both a real-world laboratory and a marquee credibility in a creative field that has been deeply wary of AI. It's also a mainstream-crossover moment: AI showing up in the culture industry as an investor and collaborator, not just as a threat in a labor dispute.

The honest caveats: commenters were quick to be skeptical of the 'not for training' framing, on the reasonable grounds that proximity to a studio's films and creative process is itself valuable to an AI company, whatever the contract says -- and the public can't see the contract. The official position is clear; whether the practical reality stays cleanly on the tooling side of the line is something only time will show. And like any splashy partnership, the announcement is easy; the test is whether real, useful tools come out of it, or whether it ends up as a prestige association that produces more press than product. For now it's a genuine, multi-outlet-confirmed deal -- and a notable vote of confidence that AI's future in film is collaborative, at least on paper.


Primary source, verified: read the paper →