Trust & safety

Moderation that runs before we spend a cent

Screening happens at the prompt stage, before the render is paid for, and again on the finished clip. Both stages are logged, and edge cases go to a person rather than straight to a library.

Categorically blocked

Sexual content involving real people

Any sexual or intimate depiction of a real, identifiable person is blocked outright. There is no consent workflow that unlocks this — the service does not produce sexual content at all.

Anything involving minors

Prompts combining minors with sexual, suggestive, or exploitative framing are blocked and logged. Reports that meet the legal threshold are escalated to the appropriate authorities and NCMEC.

Impersonation & harassment

Face swaps, deepfakes, and content that harasses, degrades, or defames a real person are blocked at the prompt stage.

Election manipulation

Depictions of real candidates, officials, or voting processes presented as real footage are blocked, regardless of intent or claimed satire.

Held for human review

Not every risky prompt is an obvious violation. These go to a moderator queue and are not released to the requester until a person signs off.

  • Prompts that name or clearly describe a real public figure in a non-sexual, non-electoral context
  • Prompts referencing minors in an ordinary, non-sexual context
  • Recognizable trademarked characters, logos, or brand assets
  • Output the classifier scores as ambiguous even when the prompt looked clean

Enforcement

Strike 1

The render is blocked, the reason is shown, and the attempt is logged.

Strike 2

Blocked again with a formal warning that the next violation suspends the account.

Strike 3

The account is suspended. Remaining credits are forfeited under the Terms.

Severe violations — anything involving minors, or a credible threat — skip the strike ladder and result in immediate suspension.

Something got through?

No filter is perfect. If you find content generated on FlowPixel AI that depicts you, infringes your copyright, or breaks our policy, tell us and we will action it. Reports are triaged within one business day.