AI ads and honesty: where I draw the line

AI can now make an ad with a voice, a face and a script that never existed a minute earlier. That is useful, and it is also a responsibility. When people ask me about AI generated ads ethics, I don't start with what the technology can do. I start with a simpler question: would the viewer feel cheated if they knew how this was made?

I build Bolo Media, which helps small and local businesses make video ads, including talking-presenter (UGC-style) ads, with AI. So this is not an abstract debate for me. These are my personal principles, the lines I think a builder in this space should hold. They are how I think, not a legal document.

Why AI generated ads ethics matters more for small businesses

Big brands have reputations built over decades. One awkward ad will not sink them.

A small business is different. Its reputation is the owner's name, the neighbourhood's word of mouth, the customers who come back. In many towns in Gujarat, people know the shop and they know the family. If an ad misleads them, the damage is personal, and it lasts.

So I think the bar for honesty should be higher, not lower, for small businesses using AI. The technology makes it cheap to produce an ad. It does not make trust cheap.

What I think is fair

I don't believe AI ads are dishonest by nature. An ad has always been a made thing. Studios use actors, scripts, lighting and voice artists. Nobody thinks a jingle singer personally uses the soap.

To me, these uses are fair:

• An AI voice reading the shop's own message. The words are the owner's, the offer is real, the voice is a tool for saying it clearly.
• A presenter, clearly in the role of a presenter. Someone explaining a product, the way a TV anchor or a model in a print ad would.
• Turning a real product photo into a moving video. As long as the product in the video is the product the customer will actually get.
• Short stories and local-news style formats that are obviously creative content, not pretending to be real reporting.

The common thread: the message is true, and the format does not pretend to be something it is not.

Where I draw the line

These are the things I would not want to build or help with, whatever the technology allows.

Fake customers giving fake testimonials

A presenter saying "this saree is beautiful, look at the fabric" is advertising. An AI face saying "I bought this last week and my whole family loved it" is a made-up customer telling a made-up story. The second one borrows the trust people place in real reviews. I think that crosses the line.

Using a real person's face or voice without consent

Cloning a real person, a celebrity, a local influencer, or even the owner's relative, without their clear permission is not a grey area for me. It takes something that belongs to them. Consent should be specific, informed and given freely.

Making the product look better than it is

AI can make a plain shirt look silky, a small room look large, a dish look richer than what arrives on the plate. Polishing the presentation is normal. Changing what the product is, is not. If the customer receives something different from what the ad showed, the ad lied.

Claims nobody can stand behind

"Cures joint pain in 7 days." "Guaranteed results." These are wrong whether a human or an AI writes them. AI just makes it faster to produce them, so it is worth being more careful, not less.

Pretending AI content is real news

A local-news style video is a format. It should never be used to make people believe something happened that did not.

Should AI ads say they are AI?

This is the question I find hardest, and I want to be honest that my view is still forming.

My current thinking: disclosure matters most where the viewer could be misled about something that affects their decision. If a viewer might believe a real customer is speaking, or a real event is being reported, they deserve to know. If an AI voice is simply reading a shop's price list, the fact that the voice is synthetic changes very little for the viewer.

Platforms and regulators are also moving on this, and the rules can change. Bolo has a separate, practical post on whether AI UGC ads are allowed in India for anyone who needs that side. My post here is about what I think is right, which is a different question from what is permitted.

Honesty is also good business

There is a practical side to this. Bolo has written about why a real face sells: people respond to warmth and authenticity, not just polish. An ad that feels real and is honest will usually do better over time than one that feels slick and turns out to be false.

Small businesses live on repeat customers. One customer who feels cheated tells others, and in a small town, that travels faster than any reel. I have written about why many small businesses still don't make video ads at all. If AI ads earn a reputation for being fake, that hesitation will only grow.

A line I will keep revisiting

I don't think I have every answer. The technology is changing too fast for a fixed rulebook, and I expect my own thinking to sharpen as I see more of how these ads are used.

But the starting question will stay the same. Would the viewer feel cheated if they knew how this was made? If the honest answer is yes, the ad should change. That is a simple test, and so far it has been a good one.