Google drops $75 million into A24 for DeepMind-backed AI filmmaker tools

Google just put $75 million into A24 to build an AI partnership with DeepMind, and the most important detail is also the most boring-sounding: Google says it will not get access to A24’s archive or data. That single constraint tells you what this deal is really about. It is not about training on A24’s library. It is about shaping new “tools of the future” for creators by working alongside the people who make the finished product.

According to the reporting cited in The A.V. Club, Google’s investment is tied to work where A24 collaborates with DeepMind to develop new technologies to assist filmmakers, with the Wall Street Journal and Variety both framing it as R&D that spans multiple projects over time. Google’s announcement language is classic for these partnerships: “creates a deep research and development collaboration” across “multiple projects over time,” designed to “help artists develop new workflows and techniques,” and “ensures the tools of the future are shaped by the creators who use them.” In other words, the promise is user-centered tooling, not raw content ingestion.

From an executive perspective, this is where the deal gets interesting, because it sits at the intersection of two incentives that are starting to collide. On one side, tech companies want AI deployed into real-world workflows where usage sticks. On the other, content companies have obvious data and IP reasons to be cautious about who sees what and what gets used to train models. The explicit statement that the deal “does not give Google access to A24’s archive or data” is a guardrail, but it is also a reminder that data access is the negotiation battleground in AI deals. If you are on the governance side of a studio, you are not just asking “will this improve production?” You are asking “what exactly are we allowing, and what are we protecting?”

A24 partner Scott Belsky added more color in an interview, suggesting the tools “won’t look anything like the prompted generation type of AI that people feel uncomfortable with.” That framing matters because it is an attempt to separate “assistive” AI from the kind of output people associate with uncanny, low-trust generation. Even if you buy the distinction, it still raises the question of what “behind the scenes” means in practice. The source points to A24’s earlier dabbling, including The Brutalist, which used AI technology to clean up Adrien Brody’s Hungarian accent. That is a good example of AI positioned as refinement rather than replacement, and it aligns with Belsky’s notion of tooling that supports the creative pipeline.

The industry pressure is not purely technical. It is cultural and labor-related, and the A.V. Club source connects the current wave of AI support to discomfort among creators and artists. Director Martin Scorsese recently gave his stamp of approval to AI for making storyboards, which matters because storyboarding is the kind of early-stage labor that audiences rarely think about. If AI accelerates or changes that workflow, it can shift demand for certain skill sets even before a film hits principal photography. Kane Parsons, the 21-year-old director who recently directed A24’s biggest hit ever, goes further in highlighting “genuinely harmful consequences” of AI, emphasizing that even if AI is sometimes “just tools, not wholesale generation,” the creative experience can be compromised. In the reporting cited, Parsons tells The Australian: “Creatively, I get no enjoyment from using those tools,” and “It defeats the purpose entirely for me.” Whether you agree with that view or not, it puts a spotlight on the real risk studios manage when adopting AI: adoption can create internal friction, not just external PR.

Zoom out and you see why this deal is a “signal” for the wider market. The source opens by noting a repeated pattern: more companies that have already invested heavily in AI are insisting AI needs to be part of filmmaking. The investment is $75 million, and that number is not small enough to be treated like a lab experiment. It suggests studios are becoming downstream partners in AI R&D, not just customers. It also suggests boards and executive teams will increasingly face these questions as recurring agenda items: do we partner to influence the technology direction, or do we resist and risk falling behind on workflow capability? Google’s approach, at least in this announcement, tries to answer the “influence” part by tying tooling development to creators who use it.

Regulatory and policy pressure is part of why that “no archive access” detail lands. Even when regulators have not caught every edge case, the direction of travel is toward clearer rules on data rights, consent, and training. Deals like this can become templates: they demonstrate a structure where funding and collaboration happen without opening proprietary libraries. That could reduce legal exposure, but it can also set expectations. If creator studios want similar guardrails, they may start demanding them as standard terms, and AI vendors will have to compete on governance as much as performance.

Strategically, peers inside media and creative businesses should treat this as a board-level briefing, not just an entertainment headline. When a major AI player invests in a studio while limiting data access, it frames a model where collaboration is allowed, but ownership boundaries remain. The second-order implication is that studios may be pressured to operationalize “creator-shaped tooling” while still managing talent concerns about what AI changes. If you lead a creative business, the real competitive edge may not be whether AI exists in your workflow. It may be whether you can adopt AI in a way that protects your assets, preserves creative intent, and prevents internal backlash from turning experimentation into a reputational slow burn.