AI in Post-Production: Why Studios Are Investing and What Changes for Film Crews

In March 2026, Netflix announced it had bought InterPositive, the AI company Ben Affleck founded in stealth in 2022. The price surfaced four months later in a quarterly filing: about $587 million in cash, for a team of 16. On the earnings call the day before that filing, co-CEO Ted Sarandos said generative AI workflows had been used on roughly 300 Netflix titles this year, with the largest concentration in post-production. 

That combination is the signal worth reading. Netflix, the biggest buyer so far, is putting its money for AI in the film industry into the finishing stage, where footage already exists and someone has to make it work. This guide explains what Netflix bought, why studios are paying for AI in post-production, which tasks and roles it touches, and what to check about footage and rights before any AI tool goes near a cut.

What Netflix Bought When It Paid $587 Million for InterPositive

InterPositive builds an AI model from one production's dailies, the raw footage shot each day, and filmmakers use that model in post to relight shots, reframe them, replace backgrounds and fill in missing coverage without going back to set. It does not generate scenes from a text prompt. In Netflix's announcement, Affleck described the company's job as protecting the people behind films while giving them better tools. 

Netflix paid about $587 million in cash for InterPositive, according to its second-quarter 2026 filing, as reported by Variety. The deal was announced on March 5, 2026 without a price. The 16-person team of engineers, researchers and creatives joined Netflix, and Affleck became a senior adviser. 

In Netflix's InterPositive deal, the method matters as much as the product. According to Forbes' account of Affleck's remarks at Bloomberg Screentime, InterPositive started from open video models and fine-tuned them on footage the company shot itself, so the base model learns how cinema looks before any production's dailies go in. Each production then trains its own model on its footage, and that model stays with the production. 

Netflix will offer InterPositive's AI post-production tools to its creative partners and has no plans to sell them commercially. For Netflix, AI in post-production is now an in-house capability, bought rather than built. 

Why Studio Money Is Moving Into AI Post-Production

Studios are investing in AI post-production because it cuts the time and cost of shots that are already written into a production, and because tools built on footage a studio controls avoid some of the licensing questions that general video generators raise. 

Netflix has given the clearest numbers. Sarandos said the docuseries The American Experiment carries 17 minutes of AI-enhanced footage, made twice as fast and at half the cost of other methods. That is Netflix's own account of one series, not an independently measured result. Netflix told shareholders that post teams used the tools for larger crowds, historical battle scenes and establishing shots across titles including India's Glory and Brazil's Brasil 70, and that some productions would otherwise have dropped those shots. Netflix has not defined what counts as a generative AI workflow, so the 300-title figure shows how widely the tools are used without saying how much of each title they touched. 

Netflix is not the only studio paying for AI in post-production. The deals since 2024 show where studio money for AI in the film industry stays: 

Three of the four are built around a library, a tool or footage the studio controls. The Disney agreement ended before completion because the product it rested on was discontinued, which shows the risk of tying a partnership to one product more than it judges consumer AI video. Four deals are too few to set a direction for AI in the film industry, but Netflix's is the largest, and it puts AI in finishing, on footage Netflix already has.

How AI in Post-Production Works on a Film Today

AI in post-production today repairs and extends shots inside the existing workflow: relighting, reframing, crowd and set extension, cleanup and rotoscoping, AI-assisted face tracking in the grade, de-aging and lip-sync for dubbed versions. Each AI shot still goes back to the craft that owns it for sign-off. 

Indian productions already use AI in post-production. According to The Hollywood Reporter India, the Malayalam film Rekhachithram used a de-aged composite of Mammootty built from more than a thousand archive photos, and colorist Sidharth Meer said AI was used on Jigra to track Alia Bhatt's face so her eyes could be brightened and sharpened in the grade. Bengaluru startup NeuralGarage's technology lip-synced the Telugu dub of War 2, Reuters reported, and Netflix named Glory among its titles that used generative AI this year without saying which shots. These are reported cases, not a measure of how widely AI in the film industry is used in India. 

The table describes AI post-production tools in general, not InterPositive alone. Two kinds sit behind these shots. Some are general products that an AI post-production house or VFX facility licenses and runs on many projects. Netflix's InterPositive approach builds a model for each production from that production's dailies, which is why the question of who owns that model now comes up on contracts. 

Which Roles AI in Post-Production Changes First

AI in post-production is changing tasks inside roles before it changes the roles themselves. On current reporting, frame-by-frame work is where it shows first: rotoscoping, cleanup, first-pass color matching and lining up takes. The cut, the look of the grade and the final approval of a VFX shot stay with the editor, the colorist and the supervisor. 

Practitioners interviewed by The Hollywood Reporter India expect rotoscoping and the assistant editor's job of lining up shots to be automated first, and point out that the tools still handle Indian languages poorly. Treat the table below as the direction of change. It is an emerging trend, and no dataset has measured it yet. 

Netflix's argument is that these tools add shots productions could not otherwise afford. That is an executive's claim, and no Indian dataset yet measures how AI in the film industry is changing post-production work. For where these roles sit in a career, see our guide to post-production jobs in India, and for which areas are hiring, the overview of in-demand jobs in the film industry. 

The practical effect is that the AI post-production tools you can supervise now belong on your profile next to your craft. On Artisty, the KalaAi skill check draws its questions from the skills and experience you add, so name the tools and the tasks you actually handle.

Footage, Training Data and Rights: Who Owns What

No general rule yet decides who owns an AI model trained on a film's footage; the contracts on each project do. In Netflix's InterPositive setup, as Affleck describes it, the model trained on a production's dailies stays with that production. Model ownership is only one of the questions, though. 

Rights questions in AI in post-production split into three separate things that are easy to blur. The first is rights in the footage itself. A production cannot assume it owns every input: what it may do with footage depends on its contracts with crew and cast, performer agreements, licenses for music, archive and locations, and any other rights in the frame. 

The second is permission to process that footage with AI post-production tools at all. A contract should say separately whether footage can be uploaded, whether it can be used for training, who can access it, who owns the model and its outputs, and what happens to each at wrap. For productions working with Netflix, its guidance on generative AI in production, published in August 2025, sets these terms. It is Netflix's policy for its own partners, not a legal requirement elsewhere, though a useful reference when writing these clauses. It asks partners to use tools that do not store, reuse or train on production inputs or outputs, and to get written approval before using personal data or third-party material, generating key story elements, or significantly altering a performance. 

The third layer is the performer. Changing how an actor's performance or face reads on screen needs that actor's consent; our guide to AI protections for actors in 2026 covers digital replicas, consent and personality rights in India. 

India has no settled law yet on training AI with copyrighted work. A DPIIT committee's working paper on generative AI and copyright (Part 1, December 2025) proposes a hybrid licensing model: AI developers would get a blanket license to train on lawfully accessed copyrighted works, films included, and pay royalties at government-set rates once a model earns revenue. Under that proposal, creators could not opt out. Comments closed in February 2026, and a second part on AI-generated outputs is planned. The proposal does not decide what a production can do with its footage today; the contracts on each project do. 

Before AI Touches the Cut: A Checklist for Crews and Productions

Before an AI post-production studio or tool works on a project, get written answers to eight questions about training data, storage, model ownership, consent, sign-off, languages and track record. The same list serves a producer choosing an AI post-production house and a crew member asked to run AI post-production tools. 

• What was the base model trained on, and was that footage owned or licensed? 
• Does the AI post-production house or its tool store, reuse or train on our footage or its outputs? 
• Who keeps the model trained on our dailies, and is it deleted at wrap? 
• Where does the footage sit during the work? Unreleased footage under NDA belongs in an enterprise-secured environment, not a consumer app. 
• Which shots change a performance, and is the actor's consent in writing? 
• Who signs off each AI shot, by name and craft? 
• Does the tool handle the languages the project is in? 
• Which shots has the AI post-production studio already delivered, and on what kinds of projects? 

Written answers to these eight settle who does what with the footage. They do not, on their own, settle copyright in the material, privacy or performer consent, which need their own clauses. An AI post-production studio that answers all eight clearly is easier to work with than one that only shows a reel. When a production posts a role on Artisty, the brief has a field for required software and tools, and tools can be marked as a dealbreaker, so the stack is stated before anyone applies.

Showing AI-Assisted Post Work on Your Profile

Show AI-assisted post work as specific tasks on specific projects: which shots you relit, cleaned up or extended, which tool you used, and what you supervised or approved. A production hiring for AI in post-production wants to know you can judge AI output as well as run it. 

On Artisty, every project gets its own card with your role and a description of your contribution, which is where that detail goes. Add the software you use under Tools & Softwares on the Skills tab, then take the KalaAi skill check and short interview to earn the verified badge on your profile. 

What Netflix's InterPositive Deal Tells Film Crews

The $587 million price is a bet on a narrow kind of AI: models that learn from a production's own footage and work under people who already own the shot. It is the largest example so far of studio money going into AI in post-production, though a single deal cannot show where all the investment is heading. The Disney agreement adds a narrower lesson: a partnership built around one product carries that product's risk. 

For anyone working with AI in the film industry, crews and productions alike, that points to the same two jobs. Get upload, training, model and consent terms in writing before anything is uploaded, and make the supervision of these tools visible in your credits. Add each project where you relit, extended or cleaned up shots with AI, list the tools you used, and take the KalaAi skill check on Artisty. 

Frequently Asked Questions

What is InterPositive? 

InterPositive is an AI company Ben Affleck founded in 2022 and sold to Netflix in March 2026. It builds an AI model from a production's own dailies, which filmmakers use in post to relight and reframe shots, replace backgrounds and fill in missing coverage. It does not generate scenes from text prompts, and Netflix plans to offer the tools to its creative partners only. 

How much did Netflix pay for InterPositive? 

Netflix paid about $587 million in cash for InterPositive. The Netflix InterPositive deal was announced on March 5, 2026 without a price, and the figure appeared in Netflix's quarterly filing for the period ending June 30, 2026. InterPositive's 16-person team joined Netflix, and Ben Affleck became a senior adviser to the company. 

Is AI used in post-production in Indian films? 

Yes. Reported uses include a de-aged composite on the Malayalam film Rekhachithram, AI-assisted face tracking in the grade on Jigra, and AI lip-sync on the Telugu dub of War 2. Netflix also named the Indian series Glory among titles that used generative AI in 2026. These are individual reported cases; no dataset yet shows how widely AI in post-production is used on Indian titles. 

Will AI replace video editors? 

The evidence so far points to task change rather than replacement. AI post-production tools take a first pass at syncing, logging, rough assembly and cleanup, while the cut and its pacing stay with the editor. Indian practitioners also note the tools still handle Indian languages poorly, which slows that shift on Indian-language projects. 

Can production footage be used to train AI tools? 

Only if the production's contracts allow it, and a contract should say separately whether footage can be uploaded and whether it can be used for training. Netflix's policy for its own partners asks for tools that do not store, reuse or train on production footage or outputs. In India, a 2025 government working paper proposes a licensing model for AI training, but it is a proposal, not law.