
In this review
InterPositive: Netflix's Secret Filmmaker AI Ben Affleck Built While Trashing AI
InterPositive — Fast Facts (March 6, 2026):
- What it is: An AI system that builds a model based on an existing production's dailies, then lets a filmmaker introduce that model into the post-production process to mix and color, relight shots, remove stunt wires, and add visual effects
- Founded: 2022 by Ben Affleck — in complete stealth mode until the Netflix acquisition announcement on March 5, 2026
- Acquired by: Netflix — a rare acquisition for a company that historically builds internally rather than buys . Terms undisclosed
- Team: 16 engineers, researchers, and creatives — the entire team joins Netflix; Affleck becomes Senior Advisor
- Availability: Netflix will offer access to InterPositive's tech to its creative partners and does not have plans to sell it commercially in the marketplace
- What it is NOT: Not text-prompting or generating something from nothing. Not Sora. Not a replacement for actors, writers, or directors
In January 2026, Ben Affleck sat down on The Joe Rogan Podcast and said: "AI cannot write you Shakespeare. Nothing new is created by large language models. The idea that AI is going to be making movies from whole cloth — that's bullshit."
He was already four years into building an AI company for filmmakers.
That's the InterPositive story in miniature — and it's a more interesting story than "Netflix buys AI startup." Affleck said he began focusing on the emergence of AI in 2022, admitting he was initially "scared" by the notion of computers playing a central role in production. The more he familiarized himself with what was being developed, and was alerted to some tech firms trying to "get the human out of it," he refined InterPositive's focus. He watched what Sora and Runway were building — text-to-video generation, AI-written scripts, synthetic actors — and decided to build the opposite: an AI trained on actual film footage, designed to help directors do things they already want to do, faster and cheaper, without touching the performance, the writing, or the human judgment behind the lens.
Less than a week after Netflix dropped out of the bidding for Warner Bros. Discovery, the company announced it had bought a 16-person AI startup that had been operating in complete stealth since 2022. No press releases, no product launches, no funding announcements, no website — just four years of filming on a controlled soundstage and building models. Here's what InterPositive actually does, why Hollywood is both intrigued and alarmed, and what it means for the future of filmmaking.
What InterPositive Actually Does (The Non-Sora Version of AI Filmmaking)
Every explanation of InterPositive has to start with what it doesn't do — because the default assumption for AI video tools is generative: you type a prompt, the AI generates footage. Affleck has been emphatic: "It's not about text prompting or generating something from nothing."
InterPositive works on existing footage. The workflow: filmmakers upload their dailies — the raw footage shot during production — to create a project-specific AI model trained on that film's visual language. That model then becomes available during post-production to perform tasks that currently require either expensive VFX work or re-shoots.
The model is specifically trained to understand "visual logic and editorial consistency" — the way light behaves on a specific actor's face in a specific location, the color grading choices that define a film's look, the lens characteristics that make one director's cinematography identifiable from another's. When a director wants to relight a scene, adjust a background, or generate a missing shot, InterPositive's model can do it in a way that's visually consistent with everything already shot — because it learned from that specific film's footage, not from a generic training dataset.
What InterPositive Can Do:
- Relight shots: Change the lighting on a scene in post — moving from hard midday light to golden hour, fixing continuity errors, or achieving a look that wasn't possible on the day of shooting
- Remove stunt wires: Clean wire removal without manual frame-by-frame VFX work — the model understands what should be behind the wire based on the scene context
- Fix background replacements: Replace or extend backgrounds in a way that's visually consistent with the film's existing photography — not the generic "pasted greenscreen" look
- Color and mix: Mix and color using the visual language learned from a production's own dailies
- Handle missing shots: Generate missing shots within the visual logic of the existing footage — reshoots that couldn't happen, coverage that wasn't captured, cutaways that were forgotten on the day
- Add visual effects: Add visual effects in post trained on the film's own visual fingerprint rather than a one-size-fits-all model
What InterPositive Cannot Do (By Design):
- Generate performances: The system was intentionally trained on smaller datasets centered on filmmaking techniques rather than performances, ensuring artists retain control over creative decisions
- Write scripts or dialogue: InterPositive has no language model component — it is exclusively a visual tool
- Create synthetic actors: InterPositive does not produce generative AI videos à la OpenAI's Sora — it has no mechanism for generating human faces or performances from nothing
- Work without source footage: The entire system depends on dailies from an actual production — it cannot function as a standalone content generation tool
The Stealth Origin Story: Four Years Nobody Knew
InterPositive began filming a proprietary dataset on a controlled soundstage "with all the familiarities of a full production," according to Affleck. Think of it as a film school for an AI — every production condition, lighting scenario, camera lens, and environmental challenge a working film set encounters, deliberately recreated and photographed to build a training dataset that reflects the real visual complexity of filmmaking.
The InterPositive tools do not rely on text-to-video prompts, but instead a storehouse of visual data created by human actors on a secure soundstage and used to train the company's proprietary models. This approach — using film-production-specific data rather than scraped internet video — is the technical foundation for InterPositive's claim to visual consistency. The model has never seen a YouTube clip or a TikTok video. It has seen thousands of hours of professionally lit, properly exposed, cinematographically intentional footage, shot specifically to train it.
Affleck's origin story: "In 2022, I spent a lot of time observing the early rise of AI in production. As a filmmaker, I could see how these models came up short. For artists to apply these tools towards telling the stories we dedicate our lives to, they need to be purpose-built to represent and protect all the qualities that make a great story: the nuances of filmmaking, the predictable — and unpredictable — challenges of production environments, the distortion of a lens or the way light shape-shifts across a scene."
The Affleck Contradiction: Trashing AI While Building It
This is the part of the story that will get shared. Ben Affleck has spent the last two years accumulating a public reputation as Hollywood's most articulate AI skeptic. The Joe Rogan clip went viral. During a 2024 appearance at CNBC's Delivering Alpha summit, Affleck was widely lauded for saying that AI "cannot write you Shakespeare" and that "nothing new is created" by large language models. On The Joe Rogan Podcast in January 2026: "Really, what it is, it's going to be a tool just like visual effects." The clips circulated in pro-human-creativity AI discourse as evidence that even Hollywood insiders were pushing back on the generative AI wave.
He was building InterPositive the entire time. None of those statements were false — he genuinely believes AI can't write Shakespeare, genuinely believes text-to-video generation is not the future of serious filmmaking. But they were strategically incomplete. Affleck said he was "scared" by early AI in film production, and that fear appears to be what motivated InterPositive — not opposition to AI, but conviction that if he didn't build the filmmaker-first version, someone else would build the filmmaker-last version.
"What AI is going to do is going to disintermediate the more laborious, less creative and more costly aspects of filmmaking that will allow costs to be brought down, that will lower the barrier to entry, that will allow more voices to be heard, that will make it easier for the people who want to make Good Will Huntings to go out and make it," Affleck told CNBC in 2024. The subtext he didn't say: "And I'm already building the tool that does that."
Why Netflix Bought InterPositive (And Why Now)
The acquisition is a rare move for Netflix, which historically builds internally rather than acquires. The company's last deal was Ready Player Me, an avatar creation platform, bought in December 2025.
The deal builds on Netflix management's stated position on AI: executives have stressed ways it can improve quality rather than reduce costs or staffing. Co-CEO Ted Sarandos said in 2024: "There's a better business and a bigger business in making content 10% better than it is making it 50% cheaper." InterPositive fits that framing exactly — it's a quality tool, not a cost-cutting tool. Relighting a scene in post, generating a missing shot, removing stunt wires — these are capabilities that make better films, not cheaper films. That's the narrative Netflix needed to make this acquisition politically viable inside Hollywood.
Netflix wrote in an October shareholder letter that it aims to focus on "empowering creators with a broad set of GenAI tools to help them achieve their visions." Netflix touted its use of de-aging AI in Happy Gilmore 2, and said the producers of Billionaires' Bunker used AI tools to create concept art. The Argentinian sci-fi series El Eternauta featured what Netflix described as the "very first GenAI final footage to appear on screen" in a Netflix show or film. InterPositive represents the next phase of that strategy — not one-off AI experiments on individual productions, but a platform capability available to every Netflix creative partner.
The Hollywood Problem: Why Timing Matters
The acquisition is sure to be scrutinized as it comes at the same time that above-the-line unions are embarking on a new round of talks with studios and streamers, including Netflix. SAG-AFTRA, of which Affleck is a member, is currently negotiating contract terms to include new AI-related protections for performers. The studios and ByteDance are already in litigation: some major studios have recently taken legal action against ByteDance over unauthorized use of copyrighted content in its AI video platform Seedance 2.0.
Netflix and Affleck appear acutely aware of this tension — hence the near-simultaneous video release featuring Affleck, Netflix CTO Elizabeth Stone, and Chief Content Officer Bela Bajaria together making the same argument from three different angles. Bajaria: "We believe new tools should expand creative freedom, not constrain it or replace the work of writers, directors, actors, and crews. Ben and his team at InterPositive are part of a long tradition in our industry of artists leading the way in how innovation is used in storytelling." The messaging is defensive because it needs to be — this announcement lands in the middle of active contract negotiations where every AI tool announcement is scrutinized for what it means for crew jobs.
The honest question nobody's answering:
If InterPositive can generate missing shots, relight scenes, and fix background replacements from existing dailies — how many of those tasks were previously done by below-the-line crew members: VFX artists, colorists, on-set lighting technicians who get called back for pickups? Affleck and Netflix are careful to say the tool doesn't replace writers, directors, actors, or crews. What they haven't directly addressed is which specific post-production jobs become obsolete when a director can relight a scene or generate a missing shot in an afternoon rather than scheduling a $200,000 pickup day. That question is coming in the union negotiations, and InterPositive is now exactly the kind of named tool those negotiations will reference.
Who Can Use InterPositive Right Now
The honest answer: almost nobody outside Netflix. Netflix will offer access to InterPositive's tech to its creative partners and does not have plans to sell it commercially in the marketplace. "Creative partners" means filmmakers and showrunners working directly with Netflix on productions — not independent filmmakers, not other studios, not individual users. InterPositive is not a product you can subscribe to, an API you can call, or a tool you can download.
The practical reach: Netflix will offer InterPositive to its creative community as a tool that keeps filmmakers at the center of the process. Given Netflix produces hundreds of original films and series annually across every genre and budget tier, "Netflix creative partners" is a large and varied group — from prestige drama directors to commercial genre filmmakers. The tool will likely appear as a standard post-production option for Netflix productions within 12–18 months of the acquisition closing.
InterPositive vs. Sora vs. Runway: How They're Different
| Factor | InterPositive | OpenAI Sora | Runway Gen-3 |
|---|---|---|---|
| Input | Your own production dailies | Text prompt | Text + image reference |
| Output | Modifications to existing footage | New synthetic video | New synthetic video |
| Visual consistency | Trained on your film — consistent by design | Generic style only | Style reference helps but not film-specific |
| Generates synthetic actors | ❌ By design | ✅ | ✅ |
| Use case | Professional post-production | Content creation, social media | Creative/marketing video |
| Union controversy risk | High (post-production jobs) | Very high (synthetic actors) | Very high (synthetic actors) |
| Availability | Netflix creative partners only | ChatGPT Plus/Pro users | $12–$76/month public plans |
Frequently Asked Questions
What Is InterPositive?
InterPositive is an AI filmmaking tool that builds a model based on a production's dailies, then lets filmmakers use that model in post-production to relight shots, remove stunt wires, fix missing shots, replace backgrounds, mix and color, and add visual effects — all within the visual language of the specific film, not a generic AI aesthetic. Founded by Ben Affleck in 2022, acquired by Netflix on March 5, 2026.
Did Ben Affleck Really Build an AI Company While Publicly Criticizing AI?
Yes — and he's been careful about the distinction. His criticism targets generative AI: text-prompting and generating something from nothing, synthetic actors like "Tilly Norwood," AI-written scripts. InterPositive does none of those things. Affleck's consistent public position — "AI is going to be a tool just like visual effects" — is entirely consistent with building a tool for visual effects work. The contradiction is more apparent than real once you understand what InterPositive actually does. The gap between his public statements and his private work is narrower than the headlines suggest.
Can I Use InterPositive?
Netflix does not have plans to sell InterPositive commercially in the marketplace. Access is limited to Netflix creative partners — filmmakers and showrunners working on Netflix productions. There is no consumer product, no API, no subscription tier. If you're an independent filmmaker, a student, or a small studio not working with Netflix directly, InterPositive is not currently accessible to you.
How Is InterPositive Different From Sora?
Fundamentally different at the input and philosophy level. Sora takes a text prompt and generates synthetic video from nothing. InterPositive does not produce generative AI videos à la Sora. It takes existing footage from your production, builds a model of your film's visual language from those dailies, and uses that model to modify or extend your existing footage — relighting shots, removing wires, generating missing coverage. The results stay visually consistent with what you already shot because the model learned from your footage, not from internet video.
Will InterPositive Replace VFX Artists and Film Crews?
Netflix and Affleck's stated position: no — it expands creative freedom rather than replacing crew. Bajaria said: "We believe new tools should expand creative freedom, not constrain it or replace the work of writers, directors, actors, and crews." The honest unresolved question: tasks like wire removal, color grading, relighting, and missing shot generation currently employ VFX artists and post-production specialists. If InterPositive automates those specific tasks, those specific jobs are affected regardless of the stated intent. SAG-AFTRA is currently in contract negotiations that will directly address AI tools — InterPositive will be part of that conversation.
What Has Netflix Used AI For Before This Acquisition?
Netflix has used de-aging AI in Happy Gilmore 2 and producers of Billionaires' Bunker used AI tools to create concept art. The Argentinian sci-fi series El Eternauta featured what Netflix described as the "very first GenAI final footage to appear on screen" in a Netflix production. InterPositive represents the shift from one-off experiments to a systematic, platform-level approach to AI in production — available as a standard tool for every Netflix creative partner rather than a case-by-case experiment.
Who Are the Key Netflix Executives Behind This Acquisition?
Netflix Chief Product and Technology Officer Elizabeth Stone led the acquisition from the technology side. Chief Content Officer Bela Bajaria represents the creative community's interests in the deal. Both participated in the video announcement alongside Affleck — an unusual three-way communication designed to address both the technical and creative-community dimensions simultaneously. Stone reports to co-CEOs Ted Sarandos and Greg Peters.
InterPositive (Netflix AI) Alternatives
Similar tools in Video
Reviews
Real experiences from verified users
No reviews yet
Be the first to share your experience
































































