Netflix now has its own AI video tool

Netflix’s latest AI move is a tool that edits shots using AI. This week, researchers from Netflix and Sofia University’s INSAIT released VOID, short for Video Object and Interaction Deletion, a video-editing model built to do more than cleanly erase an object from a scene. The point is to rebuild the rest of the frame so it still behaves like the real world, without broken motion, floating props, or effects that stop making sense once something has been removed.

Netflix’s own model page says the system is designed to remove an object along with the interactions it causes around it, including reflections, shadows, and other physical knock-on effects. The easiest example is, remove a person holding a guitar, and the guitar should not stay suspended in the air. It should drop.

On March 5, the company said it had acquired InterPositive, the AI film-tech company founded by Ben Affleck, which is a filmmaking technology company making AI-powered tools for movie production. In Netflix’s own announcement, Affleck said InterPositive’s first model was built to understand visual logic and editorial consistency while dealing with production headaches like missing shots and incorrect lighting. With reports saying the deal could be worth as much as $600 million. Affleck is also joining Netflix as a senior advisor.

Put next to that acquisition, VOID stops looking like a neat lab project and starts looking like a marker. Netflix is not just dabbling in generative AI on the edges of the business. It is building toward an AI-assisted post-production stack. 

VOID is public on Hugging Face and GitHub, where Netflix describes it as a model for interaction-aware video editing rather than simple object cleanup. That choice matters. It suggests Netflix wants to shape this layer of filmmaking, not just quietly use it behind the curtain.

Also read: Why OpenAI Bought TBPN and What It Means for Tech Media

Netflix’s own rules show the company knows how touchy this gets once AI moves from the lab into finished entertainment. During the company’s second-quarter investor call in July 2025, co-CEO Ted Sarandos said Netflix used generative AI to create a building-collapse VFX sequence in the Argentine series El Eternauta and said it was completed 10 times faster than it would have been with traditional tools. He also called it the first final GenAI footage to appear on screen in a Netflix original. That is the stronger signal behind VOID. 

Netflix is trying to move AI in high-end production out of the “experimental” bucket and into the normal machinery of how movies and series get made. VOID is one of the clearest signs yet of where that push is heading.