Netflix Used Generative AI on Around 300 Titles in 2026

Netflix has used generative AI at some point in the production of roughly 300 programmes and films during the first half of 2026, the streaming company disclosed in its second-quarter earnings materials released on Thursday 16 July.

The figure, revealed in Netflix’s investor letter and Securities and Exchange Commission filings dated 16 and 17 July, is the first hard number the company has attached to its use of generative AI in production. Netflix says the tools now feature across its pipeline, from concept development and pre-visualisation through to post-production and release.

Crucially, this is not Netflix saying it is making complete episodes from text prompts. The company’s examples point principally to work on existing footage: expanding background crowds, enhancing historical scenes, creating establishing shots where archive material is limited, and correcting visual details that might otherwise require costly reshoots.

Most of the AI activity has taken place in post-production, according to Netflix, including visual effects, editing and localisation. In its shareholder letter, the company said: “We are increasingly leveraging these tools to deliver higher quality output more quickly and at a lower cost than traditional methods.”

From stadium crowds to historical battle scenes

Netflix identified three productions as examples. Glory, an Indian crime suspense sports drama, used AI-generated stadium crowds. Brazilian football documentary miniseries Brasil 70: A Saga do Tri used generative AI to reconstruct historical visual atmosphere.

The most detailed example is The American Experiment, a five-episode US docuseries that premiered on Netflix on 24 June. The series, executive-produced by Tom Hanks and directed by Brian Knappenberger, explores the founding of the US and American democracy, with Martin Sheen voicing George Washington.

Netflix says the programme contains 17 minutes of AI-enhanced footage, used for crowds, historical battle sequences and world-building establishing shots. Co-chief executive Ted Sarandos said that material was produced twice as fast and at half the cost of previous options.

“We’re leveraging Gen AI for really complicated shots and sequences… enhancing crowds, or historical battle scenes, those kind of things,” Sarandos said on the company’s earnings call.

He added that some productions would otherwise have omitted shots they could not afford. Netflix’s shareholder letter made the same case, stating: “In some cases, productions would have had to leave out key shots and sequences in the absence of GenAI technology.”

AI is moving beyond production

Netflix is also applying large language models and AI within its app. The company says it intends to use them to improve title discovery and better understand member preferences, while management pointed to work on personalisation and search alongside the production tools.

The scale of the production disclosure is notable. Netflix released 597 original titles in 2025, after 589 in 2024 and 568 in 2023. If its 2026 output lands in a similar range, the roughly 300 AI-assisted productions identified so far could mean the technology has touched more than half of Netflix’s originals this year.

The company’s AI infrastructure now includes Eyeline, its visual-effects research group; INKubator, an in-house studio focused on AI-assisted animated short-form content; and InterPositive, the AI filmmaking startup Netflix acquired in March 2026. InterPositive’s tools work directly with existing production footage, Netflix says, rather than operating as simple text-to-video generators. Ben Affleck, InterPositive’s co-founder, is now a senior adviser.

Netflix paid $587 million in cash for InterPositive, with performance-linked earnouts that could bring the total to $600 million. The startup’s patent applications claim its tools can reduce visual-effects costs by around 50% and spending on background actors and stand-ins by roughly 70%, though those are claims rather than independently verified results.

No title list and no viewer labels

Netflix has not published a complete list of the approximately 300 affected titles. It has also not added on-screen disclosures to productions that use generative AI, and there is no current industry standard requiring it to do so.

That leaves subscribers with no straightforward way to establish whether a scene they are watching was algorithmically built or altered. The disclosure currently sits in investor materials rather than in programme credits or Netflix’s viewing interface.

Netflix has set guardrails for its production partners. Intended AI use must be disclosed to Netflix, while final footage, talent likenesses, personal data and third-party intellectual property can require written approval. Its guidelines prohibit generating or replacing performances and union-covered work without consent and appropriate agreements.

The company has framed the technology as an aid to creatives rather than a replacement for them. “It takes great artists to make something great, and AI is not changing that,” Sarandos said. “AI will give creatives better tools to bring their visions to life.”

Netflix spent $17.1 billion on content in 2025, up from $16.2 billion in 2024, and could spend as much as $20 billion this year. Sarandos expects savings from AI-assisted workflows to be reinvested in more programming. Whether that produces more ambitious scenes, lower production costs, or greater pressure on post-production workers will be one of the central questions as Netflix expands the technology’s role.