Should a salon post AI-generated hair photos?

Not as your own work. Nothing stops you from posting one, and the platforms treat it as a different kind of picture than the one you took. Meta requires disclosure on generated or altered photorealistic video and realistic-sounding audio, and says it may apply penalties when people do not.^[1] TikTok has required creators to disclose AI-generated content for years, and since May 2024 it reads the metadata that generating tools write into a file so it can apply the label itself.^[3][4] The label is the small cost. The large one is that a generated photo is not a result anybody can ask you for.

What the platforms do with a generated image

Meta's disclosure requirement covers video and audio. A still generated from a prompt is not named in it. What Meta says instead is that it applies its AI info label "to a wider range of video, audio and image content when we detect industry standard AI image indicators or when people disclose that they're uploading AI-generated content."^[2] For a still, the mechanism is detection rather than a rule you agree to follow.

TikTok arrived at the same place from the other direction. In May 2024 it began automatically labelling uploads that already carry a Content Credentials tag, the C2PA metadata that generating tools attach to a file, which catches images made somewhere else and brought in.^[3] DALL-E 3 was among the tools named at launch, and its own rules already required creators to disclose AI-generated content, so the automatic label sits underneath a requirement that was there first.^[4]

Edited is not the same as generated

Straightening a crooked mirror shot, cropping tighter, fixing exposure, and cloning out the towel on the floor are not what any of the labelling targets. Meta drew that line in September 2024: for content it detects was only modified or edited by AI tools, the AI info label moved into the post's menu rather than sitting on the post, and the visible label was kept for content detected as generated by an AI tool.^[2] The retouch does not put a badge on your work. The prompt does.

The signal also travels in the file rather than in how the picture looks. Whether it looks real was never the test.

Why the label is the smaller problem

A label costs a line of grey text. Meta's stated approach is to keep the content up so it can add context to it, unless the content otherwise violates its policies.^[2] What the generated photo costs is the thing the post was for.

A booking starts when somebody sees a specific result and can ask for it. The honey balayage on the second row of the grid is a thing a stranger can point at and say: that, on me. A generated image has no appointment behind it. If it does bring somebody in, the consultation starts from hair that never grew on a head, under light that was rendered rather than lit.

The before and after is the format that carries the most weight in this industry, and it rests entirely on the two frames being the same person. Shooting a before and after that reads at thumbnail size covers the framing. The premise underneath it is that both photos happened.

The number I am not going to give you

There is a figure going around about how much of the inspiration-photo pile is now AI-generated, quoted with a percentage on it. I could not trace it to a study today. Every version pointed at another marketing post, which pointed at nothing. So this post carries no number, and for what it is worth, anyone quoting one is worth asking where they got it.

What the generated photo was standing in for

An AI image usually goes up because the camera roll did not have the shot. That gap comes in three shapes: nobody photographed that service, the photo exists and is unusable, or the work is all there and nothing ever got made into a post.

Each has a cheaper fix than generating a picture. If the service was never photographed, the capture guide lists what is worth grabbing while the phone is already in your hand. If the photo exists and is unusable, that is almost always framing and light rather than the camera. The third one is the real one, and it is not a discipline problem.

Rebecca is a barber in Indianapolis and she is very good at it. She was also invisible online for years, and not because she had nothing to show. The sorting and the picking and the writing land at nine at night after a full day of clients, and at nine at night they do not happen.

I built SalonFawn for that gap, so grain of salt. It takes the photos already on the phone and turns them into posted content: captions written from what is actually in the photo rather than from the category, deduplication that catches near-duplicate shots so the same look does not post twice, and a per-photo toggle that keeps specific images out of the rotation.

The limits at the same volume. The AI acts as an editor and a scheduler, never a generator. It never invents facts about services, prices, or availability, and it never generates footage. Video is cut from clips you filmed, and every video holds for your approval before it goes anywhere. That also means it cannot fill a hole in your camera roll. If the shot was never taken there is nothing to post.

Photos are the entry point and video is an upgrade path, so a stylist who never shoots video can use it fully. Fourteen days free, cancel anytime. Your call.