Cover or erase? Hiding vs removing people in photos

When someone is in a photo you want to share but they should not be identifiable, you have two very different options: cover them (blur, pixelate, a solid cover, or an emoji) or erase them entirely with a generative “remove” tool. They look like they solve the same problem. They do not. One hides part of a real photo; the other invents a new version of the scene.

Here is how to choose.

The honesty problem

We are getting worse at telling real from edited. In a September 2025 survey, 76% of Americans said it is very or extremely important to be able to tell whether images are AI- or human-made — and 53% said they were not confident they could tell (Pew Research Center, Sept 2025). That gap matters. A covered face is obviously covered: the viewer knows something was hidden and that the rest of the moment is intact. An erased person leaves no trace, so the photo silently claims a different reality.

The scale of generative editing is why platforms are reacting at all. Adobe’s Firefly alone generated more than 22 billion assets in under two years (Adobe Blog, Apr 2025), and the Content Authenticity Initiative — the industry effort to attach provenance to images — passed 5,000 members, including Adobe, Sony, Canon, and Nikon (CAI, Aug 2025). Editing reality is now common enough that the ecosystem is building labels for it.

Cover vs erase, side by side

	Cover (blur · pixelate · solid · emoji)	Erase (generative remove)
Alters reality?	No — hides part of a real photo; the moment stays true	Yes — invents new pixels to replace the person
Honesty	Visibly signals something was hidden	Silently rewrites the scene; viewer can’t tell
Speed	Fast — cover every face at once, no cleanup	Slower per subject, plus fixing artifacts
Look / artifacts	Clean, intentional mark	Can warp backgrounds, leave smears or “ghosts”
Privacy	On-device; cover baked into the export, no hidden original	Varies; the original is still rewritten, not just hidden
Platform AI-label risk	Low — blur and filters are generally exempt	Higher — reality-altering edits can attract a label
When to use	Consent, privacy, honesty; people who should be hidden but present	True cleanup: a stray distant bystander or object

What the platforms actually do

This is where covering quietly wins. YouTube now requires creators to disclose realistic altered or synthetic media — but it explicitly exempts filters, beauty effects, color and lighting adjustments, and background blur from that requirement (YouTube Official Blog, Mar 2024). In other words, covering sits squarely in the exempt category, while fabricating a scene is the kind of edit that needs disclosing.

Meta goes further on the labeling side, tagging AI-edited images with “AI Info” or “Made with AI” markers based on C2PA Content Credentials (Meta Newsroom, Apr 2024). A generative erase can carry that signal; a blur or solid cover is not a fabricated scene and does not.

When each one is the right call

Cover when honesty and consent matter — which is most of the time you are hiding a person:

• Strangers and bystanders who did not agree to be posted. See whether you should cover strangers’ faces.
• Group shots and event crowds where you want everyone hidden except a consenting few. Here is how to cover a whole crowd at once.
• Any time the photo should stay a true record of the moment, just with identities protected.

Erase for genuine cleanup — a single stray person far in the background, or a piece of litter — where the goal is a tidier shot and you accept that you are changing what the photo depicts.

Where Poof fits

Poof is a cover tool, full stop. It finds every face in your photo on your device and covers them at once, and you can tap to drop a soft, auto-sized cover on any spot it missed. You pick the style — emoji, blur, pixelate, or a solid cover — and tune the coarseness; if you are deciding between them, see blur vs pixelate vs solid. The cover bakes into the exported pixels with no hidden original underneath, so it is not practically recoverable later — and it never pretends the person was not there.

Poof does not erase anyone. The built-in erase/clean-up tool in your photo app is the place for generative removal, with the trade-offs in the table above. For the bigger question of when to hide a face at all, start with whether you should cover strangers’ faces.

The bottom line

Covering keeps the photo honest, fast, and private; erasing rewrites reality, can look fake, and may earn an “AI-edited” label. When you are hiding a person — for privacy or consent — covering is almost always the right tool. Save generative erase for actual cleanup.

Want covering to be the easy default? See what Poof can do.