CalMatters Tests AI Video Models on Dance—None Could Actually Dance
CalMatters and The Markup just published rigorous testing of four leading AI video generation models—OpenAI's Sora 2, Google's Veo 3.1, Kuaishou's...
Netflix has acquired InterPositive, the filmmaking AI company founded by Ben Affleck in 2022, bringing the entire team into Netflix and adding Affleck as Senior Advisor. The stated mission: AI tools built by filmmakers, for filmmakers, designed to protect creative intent rather than automate it away.
The announcement is heavy on principle and light on product specifics, which is worth noting. But the underlying story — a working director building proprietary AI tools from a controlled soundstage up, trained on purpose-built datasets rather than scraped internet content — is substantively different from most AI-in-Hollywood narratives, and deserves to be read as such.
Affleck's origin story is more technical than the press release framing suggests. Starting in 2022, he assembled a team of engineers, researchers, and creatives and filmed a proprietary dataset on a controlled soundstage — a full production environment designed to capture the actual vocabulary of cinematographers and directors. The goal was not a general-purpose image model. It was a model trained to understand visual logic and editorial consistency under real production conditions: missing shots, background replacements, and incorrect lighting.
The first model was deliberately narrow. It focused on filmmaking techniques rather than performances — a distinction that carries significant ethical and legal weight in an industry where actor likeness rights are actively contested. The tools were built with restraints designed to keep creative decisions with artists and prevent the technology from overriding the story it's meant to serve.
That's a meaningful design philosophy, not just marketing language. Whether it survives contact with Netflix's scale is the genuinely open question.
Netflix has been consistent in its public position that AI should serve storytellers rather than replace them. Whether that position holds under the pressure of production economics is something the industry watches carefully. This acquisition, at minimum, gives Netflix a team and a toolset built from a creator-first premise, with Affleck providing both technical credibility and significant cultural cover in an industry that has been vocally hostile to AI encroachment.
From a strategic standpoint, owning proprietary AI infrastructure for production — rather than licensing general-purpose models — gives Netflix more control over how the technology behaves, how it's governed, and how it's communicated to talent and guilds. After the WGA and SAG-AFTRA strikes of 2023, in which AI protections were central negotiating points, that governance dimension is not incidental. It's core to Netflix's ability to work with the creative community at all.
The acquisition also positions Netflix ahead of competitors who are either avoiding the AI conversation publicly or deploying general-purpose tools without filmmaker-specific design. Purpose-built, ethically grounded AI infrastructure is a competitive differentiator in talent relationships, not just in production efficiency.
The announcement is notably quiet on several things that matter. There's no disclosure of what InterPositive's tools will actually do within Netflix productions, when they'll be deployed, or how the company plans to share access with the broader creative community — something Affleck gestures toward but doesn't specify. There's no detail on how the technology interacts with existing union agreements or on what consultation took place with guilds prior to the acquisition.
The commitment to making these tools available to the wider filmmaking community is stated as a value rather than a plan. That gap between principle and mechanism is where most AI ethics commitments in entertainment have historically dissolved.
For content and brand teams watching how AI integrates into creative industries, the InterPositive model offers a framework worth studying: narrow datasets, purpose-built training, human oversight built into the architecture, and a domain expert at the founding table. That's the opposite of deploying a general-purpose model on a specialized creative problem and hoping it performs.
The lesson for marketing teams building AI into content workflows is the same one Affleck applied to filmmaking: the tools that earn trust are those designed with the craft's specific requirements in mind, not those that promise to do everything. Generality is not a feature when your output needs to be specific, controlled, and defensible.
Netflix and InterPositive have built a credible foundation. What they build on it is still to be determined.
If you want to integrate AI into your creative and content operations in ways that actually hold up — technically and reputationally — Winsome Marketing's strategists can help you build something worth standing behind.
CalMatters and The Markup just published rigorous testing of four leading AI video generation models—OpenAI's Sora 2, Google's Veo 3.1, Kuaishou's...
Communities across the United States are hitting pause on data center construction. New York State has proposed a three-year moratorium on new...
In 2023, Zoom quietly wrote a $51 million check to a scrappy AI lab called Anthropic. This week, that bet looks like it could be worth somewhere...