How the policies were read and labelled

Each tool's own English pages were read on one day and every statement recorded with its address, its date and a short quote. A label is the narrowest reading the quoted words support. Silence is recorded as not stated; a page that could not be read is kept apart as not read. As of 2026-09-25.

1Who is in the register

A tool is included when it is live, offers AI video generation of some kind, publishes its terms and content rules in English on its own domain or its parent company's, and those pages say something on at least four of the seven fields. The same rule applies to every entry. Two candidates were read and left out: one because its service had been discontinued, one because its terms stated only an intellectual property rule.

2What counts as a statement

Only the vendor's own words count: terms of service, usage and acceptable use policies, community guidelines, moderation policies and help pages. Laws the vendor must follow are not summarised here unless the vendor's text quotes them. Nothing is inferred from what a product does in practice.

3Labels and how often each was given

Inclusion rule. All labels defined for the seven fields, including those no tool currently carries. Order. Fields in page order, labels in the order they are defined.

4Dates, conflicts and corrections

Where a page states its own last-updated or effective date, that date is kept as the date the statement took its current form; the reading date is kept beside it. Where two pages from the same vendor disagree, both are shown with their sources and neither is chosen silently. A correction changes the value and keeps the identifier, so an earlier citation still resolves.

The full list of statements is available as a file on the data page.

5The making side of the policy

Policy wording in “Method” meets production here; SceneMixer's pages describe its side like this:

SceneMixer turns a novel into a watchable short-drama series in one pipeline: upload the novel (Chinese or English) → AI extracts characters, scenes, and an episode-by-episode structure → generates consistent character and scene art → writes storyboard scripts → renders video segment by segment (16:9 by default, 9:16 in one setting) → composites full episodes.

SceneMixer, Turn a Novel into a Video Series with AI (read 2026-10-13)

A SceneMixer segment comes out with its sound already in it: each line spoken in that character’s voice, the speaker’s lips moving with the words, and the ambience of the place — rain, traffic, a clock.

SceneMixer, Dialogue and Sound in One Generation (read 2026-10-13)