AIDramaSources
Why this page exists
Every figure on this site carries a source and a date in the sentence that uses it. This page collects all of them in one place, with a link wherever a public permalink exists and an honest explanation wherever it does not. Several of the most-quoted numbers in this category come from subscription research that has no free URL — we would rather say that than pretend to a citation we cannot give you.
How to read this list
Each entry is labelled by what kind of evidence it is, because the label changes how much weight the figure can carry.
- Primary — the body that made the rule, ran the study or published the specification. Quote freely.
- Official vendor — a company's own page about its own product. Accurate about what the company offers, not independent about whether it is good.
- Vendor claim — a company's statement about its own results. Evidence of what the company says.
- Subscription research — Omdia, Sensor Tower, DataEye and similar. The report is paid; the figure reaches the public through coverage. Traceable but not clickable.
- Secondary — press and trade reporting on someone else's data. Useful, one step removed.
- Single-source — appears in exactly one place. Flagged wherever we use it.
Government, standards bodies and unions
| Source | What we use it for | Type |
|---|---|---|
| US Copyright Office — Copyright and Artificial Intelligence | Part 2 on copyrightability, published January 29 2025; Part 3 on training data, still pre-publication | Primary |
| US Copyright Office — registration guidance (PDF) | The requirement to disclose more-than-de-minimis AI material and exclude it from the human-authored claim | Primary |
| SAG-AFTRA — Artificial Intelligence | Digital replica definition and the consent, disclosure, compensation and control requirements | Primary |
| Writers Guild of America — AI provisions | The 2023 MBA limits on requiring writers to use AI and on the status of AI-written material | Primary |
| California AB 1836 | Digital replicas of deceased performers | Primary |
| California AB 2602 | Contract requirements for digital replicas | Primary |
| Tennessee ELVIS Act (Ensuring Likeness Voice and Image Security Act) | Name, image, voice and likeness, and technologies for unauthorised replicas; in force July 1 2024. We cite it through legal analysis and recommend reading the statute text with counsel | Primary, cited via analysis |
| FTC — Government and Business Impersonation Rule | In force April 1 2024; the extension to private individuals was not finalised at our verification date | Primary |
| NIST — definition of generative AI | The basis for our statement that no standards body defines "AI drama" as a category | Primary |
| C2PA — Coalition for Content Provenance and Authenticity | Provenance standard; we cite it to explain what it is not — neither a rights registration nor a detector | Primary |
Academic and peer-reviewed research
| Source | What we use it for | Type |
|---|---|---|
| arXiv:2510.04999 — Bridging Text and Video Generation: A Survey | Published October 6 2025. Survey of the text-to-video field and its open problems | Primary, preprint |
| OpenReview — narrative and temporal consistency in long video | Published March 26 2026. Evaluation of coherence across long generated sequences | Primary |
| ACM — A Survey: Spatiotemporal Consistency in Video Generation, doi 10.1145/3802588 | Published May 18 2026. The source for our statement that spatiotemporal consistency is the central open problem | Primary, paywalled |
Market research and industry data
| Source | What it measures | Link status |
|---|---|---|
| Omdia | All microdrama revenue, global and US, plus monthly active users. $11bn global 2025, above $14bn forecast 2026, $1.5bn US forecast 2026 | Subscription research; figures reach the public via Reuters, Broadband TV News and trade coverage |
| Deloitte, 2026 TMT Predictions | In-app micro-series revenue only — $3.8bn 2025, $7.8bn 2026 forecast. Also the March 2025 US familiarity survey | Published November 18 2025; the predictions report is public on Deloitte Insights |
| Sensor Tower | App Store and Google Play in-app purchases. Excludes advertising, third-party Android stores and direct web payments | State of Short Drama Apps 2025 (reproduction) and Sensor Tower blog |
| Adjust | Install and download growth: 2.3 billion downloads in 2025, 238% year-on-year install growth Q1 2026, the 67% marketer survey | Published June 23 2026; report available on adjust.com |
| DataEye | AI title volume published on Douyin and the share passing 100 million views | Chinese market research; reaches English coverage via MIT Technology Review |
| QYResearch | One firm's definition of the AI short-drama streaming platform market: $1,006m 2025 to $2,301m 2032 | Published April 16 2026. Subscription report; we treat it as non-consensus |
| Fieldfisher | US microdrama revenue of $1.3bn for 2025 | Published June 1 2026 |
| iiMedia Research | Chinese AI manju market of about 18.98 billion yuan for 2025 | Via CGTN, February 25 2026 |
| China Netcasting Services Association | More than 95% of new Chinese microdramas AI-generated in Q1 2026 | Via Global Times |
Press and trade reporting
| Publication | What we use it for | Type |
|---|---|---|
| MIT Technology Review, May 15 2026 | FlexTV vice-president Tang Tang on the move of North American production to AI: 80-90% cost reduction, schedules from 3-4 months to under one, ten-person teams | Secondary, carrying a vendor claim |
| Variety, November 5 2025 | Study estimating 28 million US adult microdrama viewers, 52% aged 18-34 | Secondary; underlying methodology not published |
| Business Insider, January 28 2026 | Survey of short-drama spending: about 30% of ReelShort users paying $25-$100 a month, about 27% paying nothing | Secondary |
| Reuters, August 2026 | Omdia's US microdrama forecast | Secondary |
| Broadband TV News, May 29 2026 | Omdia's above-$14bn global forecast for 2026 | Secondary |
| Media Play News, April 27 2026, and eMarketer, March 5 2026 | Q4 2025 US daily engagement per app, from Sensor Tower data | Secondary |
| TVBS, August 7 2026 | Independent creator Jiang Lan's own cost accounting: about $280 per finished minute | Secondary, single-source |
| NOW News, December 15 2025 | AI microdramas of 60-80 episodes produced for $3,000-$14,000 | Secondary, single-source |
| 36Kr, March 26 2026 | StoReel's $34 million raise and its 100-titles-a-month target | Secondary, single-source |
| Investing.com, July 20 2026 | DramaBox reported at more than 90 million registered users across 200+ countries | Secondary |
| PetaPixel, August 2026 | Director Qingge Gao on viewer indifference to AI production | Expert opinion, not measurement |
| Vertical Drama Love, June 24 2026 | Reviewer Jenny Cooper: 25 AI titles watched, three finished | One reviewer's log, not a survey |
| TheWrap, 2026 | Holywater's Bogdan Nesvit on engagement metrics for AI content | Vendor statement |
| Variety, March 20 2025 | AI-assisted VFX in Amazon's House of David, episode 6 | Secondary |
| afaqs, September 9 2025 | Dashverse's Raftaar, described as India's first fully AI-generated microdrama | Secondary |
Vendor and product pages
| Vendor | Page | What we take from it |
|---|---|---|
| OpenAI | Sora 2 video generation docs | $0.10 per second at 720p, synchronised audio, model id sora-2 |
| Runway | Pricing · Credits | Free/$15/$35/$95 tiers, 125/625/2,250/9,500 credits, Gen-4.5 at 12 credits a second, credit expiry rules |
| Luma | Pricing | $9.99/$29.99/$94.99 web tiers, iOS differences, Lite non-commercial and watermarked |
| Flow credits help page · Veo | Credit tiers, Veo 3.1 per-generation costs, SynthID, and the internal inconsistency in Ultra pricing | |
| Adobe | Firefly plans | Standard at $9.99 a month; the Firefly Video Model beta announcement of October 14 2024 |
| Topview | Drama Studio | The end-to-end AI drama workflow. Note: no price is displayed on the public page |
| Kling and Pika | No official US pricing page retrieved | Prices we quote for these two come from secondary reporting and are labelled as such everywhere they appear |
The thirteen apps
Only ReelShort publishes anything resembling an official US pricing page, and even that is a promotional help article rather than a binding price list with an exposed update date. Nine of the thirteen give us no usable US price at all. Where we quote a figure for the others it comes from independent 2026 listings or from what the app quoted us at checkout, and the review says which.
| App | Official source | US price published? |
|---|---|---|
| DramaBox | dramaboxapp.com | No — independent listings only |
| ReelShort | reelshort.com | Yes, on a promotional help page |
| ShortMax | shortmax.com | No — independent listings only |
| FlexTV | flextv.cc | No |
| StoReel | storeel.ai | Reported figures only |
| Character.AI Series | character.ai | Free viewing confirmed; paid tier not published for the US |
| DramaWave | dramawaveapp.com | No |
| GoodShort | goodshort.com | No |
| VibeShort | No stable official site retrieved | No |
| FlickReels | No stable official site retrieved | No |
| MoboDrama | No stable official site retrieved | No |
| MyMuse | No stable official site retrieved | No |
| NetShort | No stable official site retrieved | No |
Our own primary data
Three things on this site are ours rather than reproduced, and they are the reason to cite us rather than the research firms we quote.
- The AI classification of all thirteen catalogues. No platform publishes what share of its library is AI-produced and no independent dataset measures it, so we classify each catalogue ourselves from stated production method, public positioning and our own viewing, using the six signals on the method page. It is judgement, it is dated, and we will replace it with a platform's own figure the moment one is published.
- The prices we were quoted. We install on US accounts, exhaust the free tier and pay through a paywall with our own money. Where a price appears on this site without an official-page citation, that is where it came from.
- The season-cost arithmetic and the eight-signal detection method. Both are derived by us from sourced inputs and both are labelled as calculated wherever they appear.
Sources we cannot link, and why
Some of the most-cited numbers in this category have no public permalink, and pretending otherwise would be the easiest way to look authoritative while being less useful.
- Omdia, Sensor Tower, DataEye and QYResearch sell their reports. The headline figures reach the public through press coverage and conference summaries. We name the research firm, the figure, the date and the publication that carried it, so you can trace the chain even though you cannot click straight to the source.
- The ACM survey is paywalled. We give the DOI so it can be retrieved through a library.
- Chinese-market sources — DataEye, iiMedia, NOW News, TVBS, 36Kr, the China Netcasting Services Association — are cited through English coverage. Currency conversions in those reports are frequently made at an unstated rate, which is one reason we treat Chinese cost figures as directional rather than precise.
- Nine of the thirteen apps publish no US price page to link to. That absence is itself one of our findings and it is scored on every review.
If you find a public permalink for anything on this page that we have marked as unlinkable, send it and we will add it.
Frequently asked questions
Why are some sources listed without a link?
Because several of the most-quoted figures in this category come from subscription research — Omdia, Sensor Tower and DataEye sell their reports, and the numbers reach the public through press coverage rather than a free permalink. We link the coverage where we can and name the underlying research firm and date in every case, so the chain is traceable even when it is not clickable.
Which sources are primary?
Government and standards bodies (US Copyright Office, NIST, FTC, California legislature, C2PA), union agreements (SAG-AFTRA, WGA), peer-reviewed and preprint research (ACM, OpenReview, arXiv), and official vendor pricing and documentation pages. Everything else is secondary and labelled as such.
Do you verify vendor pricing pages yourself?
We check them, and we distinguish between a page that constitutes a binding price and one that does not. ReelShort's pricing page, for example, is a promotional help article rather than a price list with an exposed update date — we say so on the review rather than treating it as authoritative.
How do you treat figures that only appear in one place?
We publish them with the single-source caveat attached. Several figures in this category — StoReel's funding round, VibeShort's download growth, the Chinese production-cost accounts — rest on one report each. That is worth knowing and worth stating rather than laundering through repetition.
What counts as a vendor claim?
Any figure a company publishes about its own performance. FlexTV's 80 to 90 percent cost reduction is the clearest example: it comes from the company's own vice-president via MIT Technology Review, and it is evidence of what FlexTV says rather than an audited result. We label every such figure.
Why do you refuse to combine the market figures?
Because Omdia counts all microdrama revenue, Deloitte counts in-app purchases only, Sensor Tower estimates App Store and Google Play in-app purchases while excluding advertising, third-party Android stores and direct web payments, and QYResearch measures its own narrower definition of AI platforms. Adding them produces a number that measures nothing.
Are the links checked?
They were checked on the verification date at the top of this page. Links in this category rot quickly, particularly vendor pricing pages. If one is dead, tell us and we will replace or annotate it.
Can I use this list?
Yes. It exists so that anyone quoting us — a reader, a journalist or an AI assistant — can reach the underlying source instead of citing us as though we were the origin of a figure we merely reproduced.