Is AI-generated video worth it for social media reach?

I tested character-swapped AI video in October 2026. One post got about 75,000 views on Facebook. Almost none on TikTok.

What does it cost?

Item	Rough cost
One 10 to 15 second clip	About $3
One video a day for a month	About $100

That's cheap enough to test for a month without much risk.

Which platform worked?

Facebook, by a mile, on that test. One post is one data point, but it told me where to put the effort first.

What kind of AI video performs?

"Where these things get exciting is when they're just past the edge of realistic but still questionable."

You don't need audio or even a point to hook people. Something slightly unbelievable stops the scroll.

Why post so many?

Because short-form is a lottery.

"98% of the videos we launch here probably will get almost no engagement. But every once in a while something will pop off."

The only affordable way to buy that many tickets is to let an agent produce and publish them for you. If a person has to make each one by hand, the math doesn't work.

Should I use AI-generated people?

I'd avoid it if you can. People are getting too mad about AI people. You can use AI in a lot of ways, but fake humans get a reaction. In ad tests they did better than I expected, but for public-facing content the backlash isn't worth it.

How would I start?

1. Budget about $100 for a month: one clip a day.
2. Post to Facebook first, since that's where my test landed.
3. Aim just past the edge of realistic. No audio or big point required.
4. Have an agent produce and publish, so volume doesn't eat your week.
5. Keep AI people out of it.

For using AI footage inside paid ads, see AI video in Meta ads.