AIDramaHome›Guides›How many AI dramas are there in the US
How Many AI Dramas Are There in the US? What Can Actually Be Counted
The short version
- There is no published count of AI dramas available in the United States, and nobody is in a position to produce one. No platform discloses how many of its titles are AI-produced, and no independent dataset measures it. What can be established is a floor: on October 3, 2026 we read 288 series titles off seven platforms, and 36 of them carry a per-title AI label.
- All 36 labelled titles sit on one app. StardustTV puts a two-character "AI" badge on individual catalogue cards. One of the 37 StardustTV titles we read, "I Don't Fight, I Expose the Emperor", does not carry it — which is what makes the badge evidence rather than decoration.
- 288 is the number of titles we read, not the number that exist. DramaBox's catalogue grid returned HTTP 403 and rendered nothing, ShortMax gave up its homepage and timed out everywhere else, and StoReel has no catalogue on the public web at all. Treating 288 as a sample of the category in proportion would be wrong.
- The upper bound is not estimable, and this page does not pretend otherwise. SagaBox and ShortPeak both state that their entire catalogues are AI-produced while labelling not one individual title, and three of the seven platforms in this capture have published nothing either way. Nothing published is never written here as no AI.
- "Rated" means five different things in this category and none of them is a count. Our six weighted axes rate an app. App-store age ratings come from the developer's own questionnaire answers. No independent benchmark rates AI series at all.
DramaBoxEditor's pick — biggest short-drama library, lowest price to go unlimited
9.8Excellent
- Thousands of series, roughly 200 new titles every month
- Free episodes daily, plus ad-unlocks — no card needed to start
- Unlimited plans from about $5.99 a week, the lowest entry price we found
- Runs on Android, iOS and in a desktop browser
Affiliate link. We may earn a commission if you install through it, at no extra cost to you — it never changes our scores or placements. How we test · Disclaimer
The direct answer
Nobody publishes a count of AI dramas available in the United States. No platform states how many of its titles are AI-produced, no regulator collects it, and no independent dataset measures it. The number does not exist in any citable form. What does exist is a floor we established ourselves: 36 individual series, read and verified on October 3, 2026, each carrying a label the platform itself published.
That is the whole honest answer, and the rest of this page is the working behind it: what is countable, how the 288 titles were read, what the 36 rest on, where each bound comes from, and why the count a reader actually wants cannot be obtained from outside the companies.
What nobody publishes
Three separate numbers would have to be public before a count were possible, and none of them is. First, a catalogue total: not one of the twenty-three apps we rank states how many series it carries. Second, an AI share of that catalogue: not one of them states that either, including the six that have published something about using AI in production. Third, an independent measurement: there is no trade body, regulator or research dataset that counts AI-produced titles on US-facing short-drama apps.
Press coverage occasionally reports a share of Chinese production, which is a different population in a different country under a different labelling regime. It cannot be multiplied by a US catalogue size that nobody has published, and this page does not do arithmetic on figures whose scopes do not match.
What is countable instead
A question with no answer still has neighbours that do have answers, and those are worth more than a guess at the original. Everything in the table below was read first-hand on a dated day, from a named place.
| What is being counted | Number | Where it comes from | Read on |
|---|---|---|---|
| Short-drama apps we rank | 23 | Our own ranking, built from apps installed and paid for on a US account | October 3, 2026 |
| Apps with a full published review on this site | 16 | Our reviews index | October 3, 2026 |
| Apps that have published something about AI in their own production | 6 | Company websites, terms pages and store listings, quoted in our disclosure tracker | October 3, 2026 |
| Apps that have published nothing either way | 12 | Same check, same day. Recorded as an absence of disclosure and nothing else | October 3, 2026 |
| Apps that mark individual titles as AI | 1 | StardustTV catalogue cards, badge text "AI" | October 3, 2026 |
| Platform catalogues we attempted to read | 7 | Our capture, build/data/series_data.json and the raw captures beside it | October 3, 2026 |
| Individual series titles we read | 288 | Same capture. Rows read, not titles in existence | October 3, 2026 |
| Distinct titles among those rows | 286 | Same capture, after removing two exact duplicates across two platforms | October 3, 2026 |
| Titles carrying a per-title AI label | 36 | Same capture. All 36 on StardustTV | October 3, 2026 |
| Titles in any platform's catalogue, AI or otherwise | not published | No platform states a catalogue total anywhere we could find | — |
| AI-produced titles available to US viewers | no published figure exists | No platform, regulator, trade body or independent dataset publishes one | — |
The last two rows are the question most people are asking. They are in the table precisely so that the empty cells are visible rather than quietly left out.
The capture: 288 titles, 7 platforms, one day
On October 3, 2026 we read 288 individual series titles off the public catalogues of seven short-drama platforms. Every title was recorded with the platform it was found on, the page it was read from and the date. The raw captures sit beside the dataset in this site's build directory, including the pages that failed and the reason each one failed.
How each catalogue was reached
Three of the seven catalogues could not be read, and the page is built around saying so rather than around the number. The table below gives what each platform yielded and, in the same row, what it withheld.
| Platform | Titles read | What was readable | What was not | Per-title AI label |
|---|---|---|---|---|
| ShortMax | 85 | Homepage only. It is server-rendered and listed its promoted rows. | Every deeper page timed out or returned 404: /dramas twice, two drama detail pages, one episode page. No detail page could be cross-checked. | None seen |
| ReelShort | 75 | Homepage and eight named shelf pages, each server-rendered. | Episode counts are not shown anywhere; the TOP and Reel Original shelves overlap and overlapping titles are recorded once. | None seen |
| SagaBox | 42 | Homepage, the /watch grid, search, and sixteen title detail pages. | The comics section and the AI character chat were not catalogued. Two reads of the homepage returned different genre strings for the same title. | None. The operator states the whole catalogue is AI-generated |
| StardustTV | 37 | Homepage, all four rows, server-rendered. | Nav items carry no href in the served HTML and /categories returns 404, so no filter or category page could be opened. Episode pages render through JavaScript. | Yes — a two-character "AI" badge on the card |
| ShortPeak | 37 | Homepage, the ranking page and fourteen title detail pages. | The ranking page is a top-20, not the catalogue. Several strings in the genre position are not genres. | None. The operator states the whole catalogue is AI-produced |
| DramaBox | 12 | Two promoted series from the homepage and ten from /resources, its own editorial index. | The catalogue grid is JavaScript-rendered and every deeper path returned HTTP 403: /drama, /dramas, /browse/266, a genre list, an episode page and two review pages. /browse returned its genre chips and no titles. | None seen, on the two pages of three that rendered |
| StoReel | 0 | Nothing. Seven URLs were requested. | The public web is a JavaScript shell: every fetch returned meta tags only, and the sitemap lists four URLs — the root, the user agreement, the privacy page and the VIP service agreement. There are no per-series pages on the public web at all. | Nothing to label |
Why 288 is not a sample of the category
288 is the number of rows we read. It is not a sample of the US AI-drama catalogue in proportion, and no share computed from it would mean anything. ShortMax contributes 85 rows from a single homepage while its detail pages timed out; DramaBox contributes 12, ten of them from its own editorial index rather than its catalogue, because the catalogue grid is JavaScript-rendered and every deeper path returned HTTP 403; StoReel contributes nothing, because its public web is a JavaScript shell whose sitemap lists four URLs.
Scale makes the point better than the caveat does. findmyromantasy.com, a third-party tracker, puts StardustTV's catalogue at 13,306 series. It publishes no capture date for that figure and we have not been able to reproduce it, so treat it as an order of magnitude rather than a measurement. We read 37. Even if the tracker is only roughly right, our StardustTV rows are a fraction of one percent of one platform. The 36 is still a floor — a floor does not need to be representative to be a floor — but nothing else in the capture can be scaled up.
The same show counted twice
Two of the 288 rows are the same series on two different apps. "A New Life" and "AI Desire System" appear on both SagaBox and ShortPeak, which share a page architecture, a coin and VIP vocabulary and the same genre filter list; a third pair differs only by a trailing word. So 288 rows are 286 distinct titles, and possibly fewer. We publish both numbers because a row count and a title count are different things, and the difference is the kind of thing that gets quietly lost when a figure is repeated.
The one hard number: 36 labelled titles
Of the 288 titles we read, 36 carry a per-title AI label, and all 36 are on StardustTV. This is the only number on this page that is a count of AI dramas rather than a count of something adjacent to them. It is a count because the platform marked each title itself and we read each mark.
What the badge is
A two-character badge reading "AI" sits on the individual title card, immediately before the poster, in StardustTV's web catalogue. It is not a section, not a filter and not a site-wide banner: there is no row, tab or category on the site named after AI, and the four homepage rows are Trending Now, New Releases, Now Streaming and Exclusive Series. The marking is per title, which is what makes it countable at all.
StardustTV also describes AI-assisted production in both of its store listings, and its Google Play listing advertises an AI-generated section inside the app that does not appear on the website. The badge and the listing are two separate pieces of evidence, recorded separately in our disclosure tracker.
The title that does not carry it
One of the 37 StardustTV titles we read, "I Don't Fight, I Expose the Emperor", has no badge. That single absence is what turns the badge from decoration into evidence. A mark that appeared on everything would say nothing about any individual title; a mark that appears on 36 of 37 is a mark the platform is making a decision about.
It is still unaudited. No published threshold says what earns the badge, nobody checks that an unbadged title was filmed rather than generated, and StardustTV could change the rule tomorrow without telling anyone. Our catalogue judgement for the app is "mixed catalogue", which is our opinion and is kept in its own column, never merged with the badge.
Why 36 is a floor and not a count
36 is the number of labelled titles we reached, on one platform, on one day. StardustTV's navigation items carry no href in the served HTML, /categories returns 404 and the episode pages render through JavaScript, so there was no way to page through the catalogue beyond the homepage rows. Whatever lies past them is unknown, and on the crawl figure above it is a great deal. Every badged title past our reach raises the floor and none of them lowers it.
The bounds, with the arithmetic
A bound without its arithmetic is a number somebody has to take on trust. Both of the floors below are written out in full, and the missing ceiling is written out as missing.
| Bound | Value | The arithmetic | What it rests on | What it cannot include |
|---|---|---|---|---|
| Lower bound on labelled AI series available in the US | 36 | 36 of the 37 StardustTV titles we read carry the platform's own "AI" badge; 36 + 0 + 0 + 0 + 0 + 0 + 0 across the seven platforms = 36 | A label the platform published on each card, which we read ourselves on October 3, 2026 | Badged titles further into StardustTV's catalogue than we reached, and any label another app adds later |
| A second, weaker floor, if company statements are accepted | 113–115 | 36 labelled + 79 rows on SagaBox (42) and ShortPeak (37), whose operators state their entire catalogues are AI-produced = 115 rows; two of those rows are the same title on both apps, so 113 distinct series | Two company statements about whole catalogues, not about any individual title | Everything on the other five platforms, and anything on SagaBox or ShortPeak we did not read |
| Upper bound on AI series available in the US | none can be computed | There is no arithmetic to show. The same three of the seven platforms in this capture (ShortMax, DramaBox and StoReel) both published nothing about AI in production and yielded no readable catalogue, so there is no total to apply a share to and no share to apply | Nothing. It is the absence of a measurement, not a large estimate | Everything |
The lower bound: 36
At least 36 AI-labelled series were available to a US viewer on October 3, 2026. The arithmetic is 36 on StardustTV and zero elsewhere, because no other platform in the capture labels a title at all. The claim is narrow on purpose: it is a bound on labelled series we read, which is the only population anyone outside these companies can verify.
A second floor, if you accept company statements
SagaBox states on its homepage that it carries AI-generated short films, and ShortPeak states in section 3 of its terms of service that its content is produced using AI generation tools. Accept both at face value and the 79 rows we read off those two apps join the 36, giving 115 rows and 113 distinct series.
This floor is weaker than the first and the page keeps them apart. Neither company labels a single individual title, so no particular one of those 79 rows can be checked by anybody; what is being trusted is a sentence about a whole catalogue. The two apps also appear to share a catalogue or a supplier, which is why the distinct figure is lower than the row figure.
The upper bound: not estimable
No upper bound can be computed, and the honest form of that sentence is a blank, not a big number. The usual route — catalogue size multiplied by AI share — needs two figures and neither is available for any platform. The same three platforms — ShortMax, DramaBox and StoReel — are both the ones that have published nothing about AI in production and the ones whose catalogues could not be read, so for those three there is no share to apply and no total to apply it to. Of the remaining four, three state something about AI in their own production and one claims human production, and not one of the four publishes a catalogue total.
And silence is not evidence. An app that has said nothing may carry a catalogue that is entirely AI-produced or entirely filmed; its silence distinguishes neither. Writing "nothing published" down as "no AI" would manufacture a ceiling out of an absence, and that is the one move this site does not make.
Why the count cannot be obtained
Five separate obstacles sit between a reader and a count, and removing any four of them would still leave no number.
No platform publishes a catalogue total
Not one of the twenty-three apps we rank states how many series it carries, in its app, on its site, in its store listing or in any press material we have read. Without a denominator, every share is unanchored and every count is a count of what somebody happened to see.
Most catalogues render only through JavaScript
DramaBox's title grid is client-side, and the paths behind it answered HTTP 403. StoReel serves meta tags and nothing else. StardustTV's episode pages render client-side even though its catalogue rows do not. ShortMax's detail pages timed out on every attempt. A catalogue that only exists after a script runs cannot be read, archived or re-checked by a reader, which also means nobody can audit anyone else's count of it.
Catalogues rotate
These apps promote, retire and re-title series continuously, and several of the rows we read sit on homepage rails whose contents are chosen for the visit. A count is therefore a count on a date, which is why every figure on this page carries October 3, 2026 and why the page will say so again when it is re-checked.
Nothing is audited
The one per-title label in the category is published by the platform that benefits from it, with no definition and no external check. There is no US rule requiring an AI label on video entertainment, neither app store requires one, and no body verifies the labels that do exist. The regulatory picture, which is the reason a US viewer sees almost no labels at all, is set out in our disclosure tracker.
The nearest thing to an independent count
Third-party metadata crawls exist and are the closest available substitute, but they count series, not AI series. The StardustTV figure cited above reports 13,306 titles and says nothing about how any of them was made, and carries no date we could check. A crawl can tell you how large a catalogue is; only a label can tell you what is in it, and labels are what the category does not have.
How AI dramas are rated
"AI drama rated" is five questions wearing one phrase, and the answers do not overlap. Nothing in any of them is a count, and only one of them is about a series rather than an app.
| What is rated | Who sets it | What it measures | Independent? | What it is not |
|---|---|---|---|---|
| Our six-axis app score | This site. The axes and weights are fixed for all 23 apps | Cost to finish a season, catalogue and refresh, production quality, free tier, device coverage, billing transparency | No — it is our editorial judgement, published with its weights | Rates the app, never an individual series |
| Apple App Store age rating | The developer answers Apple's questionnaire; Apple reviews the app | Age suitability. StardustTV is rated 12+, Playlet 17+, read from the US listings in September 2026 | No — the answers are the developer's own | Says nothing about how the series was produced |
| Google Play content rating | The developer completes the IARC questionnaire; the rating is issued from the answers | Age suitability. StardustTV and Playlet are both rated Teen | Partly — the questionnaire is standardised, the answers are self-reported | Same. Production method is not among the questions |
| Store star rating and review count | Users of the app | Satisfaction with the app, heavily mixed with billing and auto-renewal complaints | No | Not a quality measure for any series, and not a count of anything |
| Per-title AI label | The platform. Only StardustTV publishes one | Whether that one title is AI-produced, in the platform's own judgement | No — unaudited, with no published threshold for what earns it | The only per-title AI signal that exists in the category |
| An independent quality benchmark for AI series | Nobody | — | — | Does not exist. Anyone presenting one is presenting an opinion |
Our six axes
Every score on this site is the same six axes with the same fixed weights, applied to an app. They are published in full, they are identical on all twenty-three apps, and a review may not invent its own.
| Axis | Weight | What it measures |
|---|---|---|
| Cost to finish a season | 25% | What it costs in real money to watch one series to the end, by the cheapest route the app offers |
| Catalogue and refresh rate | 20% | How much there is and how fast it changes. This is the nearest axis to a count, and it is a judgement, not a number |
| Production quality | 20% | What the episodes look and sound like, AI-produced or filmed |
| Free tier | 15% | How much can be watched before any payment is required |
| Device coverage | 10% | Phone, tablet, desktop browser, television, offline download |
| Billing transparency | 10% | Whether the price is published, whether renewal is clear and whether cancelling works |
Note what is not among them: no axis rates whether a title is AI-produced, and none of them rewards or penalises an app for using AI. The disclosure question is tracked separately and kept in its own column, for the same reason the count is kept apart from the capture.
Age ratings are self-declared
An app-store age rating is produced from the developer's own answers to a questionnaire, not from anyone watching the catalogue. On the US listings we read in September 2026, StardustTV is rated 12+ on the App Store and Teen on Google Play; Playlet is 17+ and Teen. Those ratings describe claimed content suitability. Neither store asks how a series was produced, so no age rating anywhere in this category carries information about AI.
What the AI-native catalogues tag themselves
The two platforms that state their entire catalogues are AI-produced publish adult genre tags of their own. ShortPeak's published tag list includes "Adult", "Steamy", "One Night Stand" and "Forced love"; SagaBox's includes "Affair", "Age Gap", "Hidden Desire" and "Temptation", and the title strings we read off both sites read accordingly.
That is a description of how those two apps tag themselves on October 3, 2026, and it is as far as the evidence goes. We could not verify either app's store age rating from the web capture, and two apps are not the category. What this does not support is the general claim that AI-produced drama is adult drama: 36 labelled titles on a 12+-rated app sit on the other side of it.
There is no independent quality benchmark
No standardised benchmark rates AI-drama seasons, and no dataset shows what share of any app's catalogue is AI-generated. Anyone presenting either as data is presenting an opinion in a costume. Our own ratings are opinions too; the difference is that the axes, the weights and the arithmetic are published, so the opinion can be argued with.
If you came here for a number to use
Use 36, and carry its scope with it. The figure is: at least 36 individual series carrying a platform-published AI label were available to US viewers on October 3, 2026, out of 288 titles read across seven platforms. Dropping the scope turns a defensible floor into a false total within one retelling.
How to cite the figures on this page
Every number here is computed from a dated capture that is part of this site's build, not typed into the page. If you quote the 36, quote the date and the words "at least" with it. If you quote 288, quote "titles we read" rather than "titles that exist". If you need the method behind any of it, it is at how we test, and the full source list for the site is at sources we use.
What this page cannot tell you
Naming the holes is part of the answer, not a disclaimer attached to it.
- It cannot tell you how many AI dramas exist. That number is not published by anyone and cannot be derived from anything that is.
- It cannot tell you what share of any catalogue is AI-produced. No platform publishes one, including the six that acknowledge using AI at all.
- It cannot tell you that the silent apps are AI-free. Twelve of the twenty-three we rank have published nothing either way, and that is a fact about their communications rather than about their catalogues.
- It cannot be re-checked by you against the same pages. Three of the seven catalogues are unreadable without running their scripts, and none of these pages is archived.
- It will go stale. Catalogues rotate weekly and a badge can be added or dropped without an announcement. Every figure here is dated October 3, 2026 for that reason.
Frequently asked questions
How many AI dramas are there in the US?
Nobody publishes that number and nobody is in a position to. What can be established is a floor: at least 36 individual series carrying a platform-published AI label were available to US viewers on October 3, 2026. That figure comes from 288 titles we read across seven platforms that day, and every labelled one of them is on StardustTV.
Is there an official count of AI dramas anywhere?
No. No platform states how many of its titles are AI-produced, no US regulator collects it, and no independent dataset measures it. If you see a total quoted, ask what it was counted from; in this category there is nothing to count it from.
How many titles did you count yourself?
288, read off the public catalogues of seven platforms on October 3, 2026. That is a count of titles we read, not of titles that exist, and the page says so everywhere the figure appears.
How many of them are labelled as AI?
36. Each carries a two-character "AI" badge that StardustTV publishes on the individual title card. No other platform in the capture marks a title at all.
Which app labels its AI titles?
StardustTV, and only StardustTV. The badge is per title, not a section or a filter, and one of the 37 titles we read does not carry it, which is what makes it a real label rather than site-wide branding.
Does 288 mean there are only 288 AI dramas in the US?
No. 288 is the number of titles we were able to read. DramaBox's catalogue grid returned HTTP 403, ShortMax yielded only its homepage, and StoReel has no catalogue on the public web at all. A third-party tracker puts StardustTV alone at 13,306 series, undated.
Why can't you just count the catalogues?
Because no platform publishes a total, most catalogue grids render only through JavaScript, the contents rotate weekly, and nothing is audited. Any one of those makes a count hard; together they put it out of reach from outside the companies.
Could the real number be in the thousands?
It could. It could also be much smaller. There is no basis for either guess, which is why this page publishes a floor and an explicit blank where the ceiling would go rather than an estimate dressed as a finding.
How many short-drama apps are there, and how many have you reviewed?
We rank twenty-three and publish full reviews of twenty-three of them, each installed and paid for on a US account. The ranking is a list of the apps we have tested, not a census of the market.
How many apps say they use AI?
Six of the twenty-three we rank have published something about AI in their own production, and twelve have published nothing either way. Every statement is quoted with its source and date in our disclosure tracker.
If an app says nothing about AI, does that mean its catalogue is human-made?
No. Silence is a fact about a company's communications, not about its catalogue. This page never converts "nothing published" into "no AI", in either direction of a bound.
How are AI dramas rated?
Five different ways, none of them a count. We rate apps on six weighted axes. Apple and Google publish age ratings generated from the developer's own questionnaire answers. Store stars rate user satisfaction. StardustTV publishes a per-title AI label. No independent benchmark rates the quality of AI series at all.
What age rating do these apps carry?
It varies and it is self-declared. On the US listings we read in September 2026, StardustTV was 12+ on the App Store and Teen on Google Play, and Playlet was 17+ and Teen. Neither store asks how a series was produced, so the rating says nothing about AI.
Are AI dramas adult content?
Some are. The two apps that state their whole catalogues are AI-produced publish adult genre tags themselves — ShortPeak lists "Adult" and "Steamy" among its tags, SagaBox lists "Affair" and "Hidden Desire". That is two apps describing themselves, not a property of AI-produced drama: the 36 labelled titles in this capture are on an app rated 12+.
How do I tell whether the episode I am watching is AI-generated?
On StardustTV, look for the badge on the card. Everywhere else you are left with the on-screen signals, which we rank by reliability in our guide to telling whether a drama is AI-generated.
How often will these figures be updated?
Every 90 days, and sooner if a platform adds or drops a label. Catalogues rotate weekly, so treat any figure here as true on October 3, 2026 and verify the date before quoting it.
The bottom line
The count does not exist, and the most useful thing a page can do with that question is refuse to invent an answer and show what is actually establishable instead. At least 36 AI-labelled series were available to US viewers on October 3, 2026. That is a floor built from labels the platform published and we read, and it will rise as soon as anyone reaches further into the catalogue than we did.
Above the floor there is nothing but two company statements about whole catalogues, which lift a weaker floor to 113 distinct series, and then open space. The ceiling is not a large number we are declining to print; it is a measurement nobody has taken. Twelve of the twenty-three apps we rank have published nothing about AI either way, three of the seven catalogues in this capture could not be read at all, and no platform anywhere publishes a catalogue total.
If that is unsatisfying, the right address for the complaint is the platforms. Any one of them could answer this question for its own catalogue in a sentence, and the one that marks individual titles shows it costs nothing to do. Until more of them do, a floor with its arithmetic attached is the most honest thing anyone can give you, and a confident total is the clearest sign that somebody has stopped checking.
We will re-read every catalogue in this capture in 90 days, and sooner if a platform adds or drops a label. If you find a per-title AI label we have missed, send it with a link and we will verify it, recount and date the change.
Related reading
Sources
- Our own catalogue capture, October 3, 2026: 288 series titles read from the public web pages of seven platforms, with each title's platform, source page and read date recorded. Held in this site's build as build/data/series_data.json, with the per-platform raw captures — including every page that failed and the reason — in raw_big3.json, raw_sagabox_shortpeak.json and raw_stardusttv_storeel.json.
- StardustTV per-title "AI" badges, read from the catalogue cards at stardusttv.net on October 3, 2026. 36 of 37 titles carry the badge. Not linked: this site carries no direct links to the apps it ranks.
- SagaBox homepage statement that it carries AI-generated vertical short films, and ShortPeak terms of service section 3, "AI-Generated Content", last updated September 2, 2026. Both read on October 3, 2026 and quoted verbatim in our AI disclosure tracker.
- DramaBox: homepage and /resources read on October 3, 2026; /drama, /dramas, /browse/266, a genre list page, an episode page and two review pages all returned HTTP 403, and /browse returned the platform's genre chips with no titles. Recorded as unreadable, not as empty.
- ShortMax: homepage read on October 3, 2026; /dramas timed out twice, two drama detail pages and one episode page timed out or returned 404. The 85 titles are homepage rows only.
- StoReel: seven URLs requested on October 3, 2026, all returning a JavaScript shell with meta tags only. Its sitemap lists four URLs — the root, the user agreement, the privacy page and the VIP service agreement — so no per-series page exists on its public web. Zero titles recorded, and zero is not a catalogue size.
- findmyromantasy.com, a third-party tracker, puts StardustTV at 13,306 series. Read October 3, 2026; the source publishes no capture date for the figure and we could not reproduce it. It counts series, not AI series. Cited as plain text, and used only as an order of magnitude.
- US App Store and Google Play listings for StardustTV and Playlet — age ratings, read September 2026 and recorded in our reviews of StardustTV and Playlet.
- build/apps.py, this site's canonical dataset of the twenty-three ranked apps: scores, prices, our dated catalogue judgement and what each company has published. The six disclosures and the twelve silences quoted on this page are computed from it and from the disclosure register, never typed in.
- Our scoring method, the six axes and their weights: how we test. The full standing source list for the site: sources we use.