Fiction and violence in AI video policies

Fig. 1 The stages are the order this primer reads a policy in; the quotes below are examples from the register.

Most policies do not distinguish between violence in fiction and violence in the world. They name violent content, gore or threats, and the words apply to a thriller as much as to anything else. The few that do distinguish are worth knowing: one welcomes stylized violence in a fictional, non-graphic context, and one writes a general exception for artistic, documentary or educational work.

In practice the most useful line in these texts is the definition of gore. Where a vendor lists blood, dismemberment or exposed organs, a fight scene can be written to stop before them.

Extremism is the violent category named most often, usually with promote or glorify, which is a reason to keep a terrorist villain's cause off screen rather than on it.

1Examples from the register

Glorify graphic violence or gore
Adobe Generative AI User Guidelines

upload or generate unlawful, obscene, pornographic, defamatory, hateful, violent, discriminatory, harassing, terror-related, or otherwise inappropriate content
Arcloop Terms of Use

stylized violence and mature themes in a fictional, non-graphic context
Character App Content Policy

Violence or the incitement of violence.
Google Generative AI Prohibited Use Policy

Section 1
Glorifying graphic violence or gore is prohibitedBe respectful and safeAdobe Generative AI User Guidelines / since 2026-05-15 / checked 2026-09-25
Prohibited conduct
Violent and terror-related content is listed as prohibitedArcloop Terms of Use / recorded 2026-09-25
Section 1
Stylized violence in a fictional, non-graphic context is welcomedWhat we welcomeCharacter App Content Policy / since 2026-05-03 / checked 2026-09-25
Sexually explicit
Violence and incitement of violence are prohibited usesviolent, hateful or harmful activitiesGoogle Generative AI Prohibited Use Policy / since 2024-12-17 / checked 2026-09-25
Content standards
User content must not contain or promote violent materialHailuo AI Terms of Service / recorded 2026-09-25
Section 2
Avatars displaying violent or threatening imagery are prohibitedAvatarsHeyGen Moderation Policy / recorded 2026-09-25

2A generator's own account

Policy wording in “Fiction and violence in AI video policies” meets production here; SceneMixer's pages describe its side like this:

The same way as a photographic one: the character is generated once as a reference sheet and that image is sent with every shot they are in, along with a couple of face-distinguishing notes. Shots refer to the character rather than re-describing their look.
SceneMixer, AI Animated Series Generator — “How do I keep a stylised character on model across episodes?” (read 2026-10-13)

The script is written from the photos and the one-line description. In step-by-step mode you can rewrite it before anything is shot.
SceneMixer, How to Make an Ad from One Photo — “Do I need a script?” (read 2026-10-13)

3Keep reading

The primer is about reading what vendors wrote. It is not advice about what any law requires.