Useful for volume. Dangerous for truth.

It can produce variations, voiceovers, subtitles and rough cuts quickly, which genuinely helps when you need volume for testing. It does not decide what the ad should say, and fully generated footage still looks generated in ways viewers notice.

A clearly non-human presenter, yes, disclosed as such. A synthetic person presented as a real employee or customer, no. The distinction is whether the video implies something untrue about who is speaking.

Only with explicit, documented consent from the person whose voice it is. Voice cloning without consent is both an ethical and increasingly a legal problem, and no campaign timeline makes it acceptable.

For volume, substantially — twenty ad variations, multilingual voiceovers, bulk subtitling. For a single high-quality video, barely, because the expensive parts are the idea, the fact-checking and the final edit, and those remain human.

Ad platforms increasingly require it for synthetic media in certain categories, and the requirements are tightening. Beyond compliance, disclosure is simply the right position: audiences react badly to discovering it later.

It depends on the tool’s terms and the jurisdiction, and the position is genuinely unsettled. For anything you need to own outright and defend — a brand asset, a logo animation — human-made work is the safer route.