Safety and trust
What Genien checks, refuses and labels.
An ad has to be true to the product, honest about AI and fair to the people in it. Here is what Genien does about each, and what it doesn't claim to do.
Built into every run
Your product's facts, approved by you
Genien reads your page and photos, asks what they don't say, and writes a brief you approve. Every idea, frame, clip and check follows it.
Every word checked against what you may claim
The brief keeps a registry of claims your product can back up. Every concept, hook, caption and static ad is checked against it, and what breaks it is shown before you publish.
Two models judge every clip
Two different models watch each clip, sound is measured, and bad clips are redone within a budget you see first. Each video comes with a report of what was fixed and what's still off. The run log shows how that went on every test run.
AI labels on by default
A video with an AI voice or person carries a disclosure label unless you turn it off. New York has required one for synthetic performers in ads since June 2026, and California will from January 1, 2027.
Real people only with their consent
Faces and voices of real people need their documented consent, and public figures, minors and deceased people are off limits. Plans show products and hands by default. The Acceptable Use Policy has the details.
What Genien won't make
No sexual content, nothing aimed at minors, no political ads, no ads for prescription drugs, weapons, tobacco, gambling or guaranteed returns, no fake testimonials and no impersonation.
Your data stays yours
We don't use your files or the words of your prompts to train AI models, and other customers never see your brand or files. You can download your data or close your account in Settings. The Privacy Policy and the model providers page say who processes what.
Report a problem
Write to hello@genien.ai, with or without an account. We act on removal requests for intimate imagery within 48 hours, and copyright complaints follow our Copyright Policy.
What we don't claim
Trust comes from knowing the limits too.
- Genien doesn't add its own watermark or C2PA Content Credentials yet. Some model providers add theirs, and the Terms ask you not to remove them.
- The checks don't catch every mistake. Some, such as tile lines that bend, still get through, so you review each video before it runs.
- AI footage can't prove that a product works. If an ad shows or claims a result, film that shot for real.
- Nothing Genien suggests has been tested on real ad results yet, so treat its ideas as things to test.
The rules that bind both sides are in the Terms of Service and the Acceptable Use Policy.