AI short films, ranked by what they actually did

What counts as an AI short film on this page

A short qualifies when two things are true: its own title declares AI-generated visuals, and its runtime is under 30 minutes. Both halves are worth stating plainly. The declaration is a claim a video makes about itself — 694 of the 3536 tracked videos carry it, and the short end of that set is what you see here. No frame is inspected, so a genuinely AI-made short that never says so in its title is missing from this page, and a re-upload that simply claims the label would be in it.

Two exclusions are worth naming because both used to be mistakes here. A title that says "No AI" is not a declaration of AI — it is a denial, and one such trailer reached number 17 on this board before the rule was fixed. And a title that talks about AI rather than being made with it — 100 of them in the dataset, tutorials and reviews and tool launches — is tagged Talking about AI, not an AI-made film instead, which keeps them off the AI boards while leaving them on their own niche boards and searchable on the radar.

The 30-minute line is this page's own boundary, not a field in the data. The dataset's short-form bucket — Short Films & Stories — is a scraping label built from title wording and runtime, and it holds 218 entries regardless of whether they are AI-made. If you want that slice, the Short Films & Stories board is the right link; this page is the runtime-and-metadata cut, and the two numbers will not match.

"Declares AI-generated visuals" is a title-level claim, not a verdict: it is set when a video's own title says so (AI-generated visuals declared in the title), and unset when the title says the opposite or is about AI rather than made with it — tutorials, reviews and behind-the-scenes shorts are tagged separately as Talking about AI, not an AI-made film and stay off this list. No frame is inspected, and a real AI short that never says so in its title will not appear on this page.

The AI short films with the fastest view velocity

Ranked by breakout score — views per day relative to the channel's own subscriber base, weighted by whether engagement looks natural for the niche. Rank is not a quality judgement; it measures which of these shorts a small channel pushed past the audience it started with. Every row opens the short on YouTube, and the channel name goes to its page on this site. For what those audience numbers imply in money rather than reach, the YouTube channel value calculator applies your own RPM assumption to the same tracked view counts.

Top 30 of 150 tracked shorts. Median short on this list: 7 minutes long, 132 days old, 2.4K/day, 22.0× its channel's subscriber base. Page rebuilt 2026-10-11.

How short are AI short films, and does the length matter

This is the question a runtime column answers better than any list of titles. Splitting the 150 tracked shorts into three bands shows where the volume actually sits — and the age column is the part to read first, because a band that looks slow may simply be a band whose entries are older and had a longer time on the platform to accumulate their average.

One capture per video, so the views-per-day figure is a lifetime average: it cannot tell you whether a short in one band is still climbing or already flat. Treat the ordering as a direction to investigate on the videos themselves, and remember that the entry conditions mean every short here already cleared 20.0K views before it could appear — a band that looks empty is a band where nothing got that far, not a band where nothing was published.

Who is making AI short films

95 channels have at least one tracked AI short. These are the twenty with the strongest single release — a mix of solo operators and small teams, most of them publishing serialised work rather than one-off pieces.

A channel with one tracked short is a topic signal, not a repeatable format — the channel pages print each channel's tracked-video count for exactly that reason.

What repeats in the short-form end

Production markers come from the same rule-based pass as the rest of the site — title wording, runtime and market, all of it the video's own wording rather than frame analysis — and a short can carry several, so the column does not sum to 100%. Read this as what these shorts have in common at the packaging level, which is the part that travels between channels.

Markers are inferred, not credited: a short tagged with a franchise name was not verified to use it, and the counts are small enough to read as "these patterns show up", not as market share. Analysis source: rule-based inference from title + duration + market (hookType / openingSeconds / structure / repeatableChoices); provenance: api metrics + metadata only, no frame-level video analysis.

AI shorts are not only an English-language thing

120 of the 150 tracked AI shorts are English-language and the remaining 30 split across 7 other markets — the same pattern the full AI movies corpus shows, only more pronounced because shorts travel further on smaller budgets. Market here means the language a video is actually spoken in, inferred from the API's audio-language field where it exists and from title script and keyword evidence otherwise; it is an inference, not a setting the uploader chose.

AI short films: common questions

Method and limits

"Declares AI-generated visuals" is a title-level claim, not a verdict: it is set when a video's own title says so (AI-generated visuals declared in the title), and unset when the title says the opposite or is about AI rather than made with it — tutorials, reviews and behind-the-scenes shorts are tagged separately as Talking about AI, not an AI-made film and stay off this list. No frame is inspected, and a real AI short that never says so in its title will not appear on this page.

On top of that: this is a filtered dataset, not a census, so an AI short that did modestly, or one published by a channel that already got big, is absent — every entry cleared at least 20.0K views, at most 100.0K subscribers and reach of 3× its channel's subscriber base at capture. There is one capture snapshot per video, so velocity is a lifetime average and cannot show whether a short is still climbing. And the scores normalise against this dataset's own median, so they are not comparable with any other tool's numbers.

Data: YouTube Data API v3 public metrics · 3536 tracked videos overall · page rebuilt 2026-10-11 · ranking method documented on the breakout board.

Browse the short-form end

4 ways into the same corpus, each one answering a different question — which is the cheapest way to compare the watch side of AI shorts (what came out recently, what exists in languages other than English) with the make side (what repeats, which tools get named). Each slice is a real page, not a filtered view of this one; slices with fewer than 12 rows are not built at all.

• AI short films published in the last 90 days

  44 tracked shorts. Newest first, for the watch side: what the format looks like in the most recent capture.

• AI short films that come in numbered episodes

  31 tracked shorts. Numbered episodes and continuing storylines, where a channel is trading reach for a returning audience.

• AI short films that name the tool they were made with

  24 tracked shorts. The minority of shorts that say which tool made them, plus the count that proves it is a minority.

• AI short films that are not in English

  30 tracked shorts. The less crowded side of the dataset: same ranking method, different language market.

Keep going

• All AI movies and the full corpus →
• Short Films & Stories board →
• Short-form markets side by side →
• Faceless niches →
• Channel value calculator →