The question is no longer whether AI can generate video. It can. The question is what that video manages to sell, and there the answer changes a lot depending on what you are selling.

What AI already does well

  • Clean product images, backgrounds, colour variants.
  • Voiceover and subtitles, now good quality in Italian too.
  • Recuts and format adaptations from existing material.
  • Fast message tests, before investing in production.

What it cannot sell

Experience. A generated video has not used the product, does not have a home, has no reason for having bought it. In categories where the decision rests on trust, meaning supplements, beauty, services, children's products, that gap is felt and the conversions show it.

The disclosure question

Generated content presented as the testimonial of a real person is misleading, and in Europe misleading advertising is a legal problem, not an aesthetic one. If you use generated material in a way that could be mistaken for a real person, say so. The practical rule: if the viewer would feel fooled on finding out, it needs disclosing.

How to combine them without losing trust

The model that works today is hybrid: the real creator films the speaking and the product use, AI produces inserts, format variants, multilingual subtitles and versions for different channels. The face stays human, the multiplication becomes cheap.

The deciding criterion

If the video has to show that someone trusts the product, you need someone. If the video has to show how it is used or how it is made, AI can do much of the work.

Keep reading

Find UGC creators in Italy →

Frequently asked questions

Does AI replace creators?

Not in categories where trust decides.

Should AI use be disclosed?

If it could be mistaken for a real person, yes.

Where does AI help most?

Inserts, formats, subtitles, fast tests.

The best model today?

Hybrid: human face, automated multiplication.