The short version: deleting the original photo removes it from the place you control. It does nothing to the copies that other systems already made. And there are two very different kinds of "other system" that keep showing your face after you thought it was gone. Telling them apart is the whole game, because only one of them is the thing we actually remove you from.
Two systems, not one
When people say "my face is still in search," they're usually describing one of two completely separate things that happen to look identical from the outside:
A facial-recognition database
Engines like PimEyes, FaceCheck.ID, Lenso.ai, and Clearview AI crawled the web, downloaded a copy of your photo, and ran it through a model that turns your face into a string of numbers — a faceprint (also called a face embedding or template). They store that faceprint in their own database. When someone uploads a picture of you, they're not searching the live web — they're searching their own stored copy. This is what Face Privacy removes you from.
A search engine's index / cache
Google, Bing, and the like crawled the page your photo lived on and kept a pointer to it (and sometimes a cached thumbnail). This is text-and-image indexing of a public URL — not biometric matching. Getting out of this is called deindexing, and it's a different process with different rules. This is not what we do (we'll cover who handles it below).
Why deleting the original doesn't reach the database
When a face engine indexed your photo — possibly months or years ago — it made its own copy. From that moment, the engine's record of you is independent of your original. Deleting the source has no automatic effect on the copy, for three reasons:
- They kept the image, not just a link. Most face engines store a downscaled copy (or at minimum a thumbnail) of every face they scraped, precisely so they can show it back in results without re-fetching the original. Your deletion never touches their storage.
- The faceprint outlives the photo. Even if the engine deleted the actual image, the mathematical faceprint it extracted is the thing that powers matching. That template is what gets compared against new uploads. A face engine can keep matching you from the template alone — the original image is no longer needed.
- Nothing tells the engine you deleted anything. There's no signal that propagates from "I removed this Instagram post" to "PimEyes should drop its copy." The engine has no reason to recrawl a now-dead URL, and even if it does, it already has what it took.
This is why the faceprint is the part that matters. As long as your template sits in an engine's database, a brand-new photo of you that you never posted anywhere — someone else's group shot, a tagged event photo — can still match you. The data point that identifies you isn't any one picture. It's the faceprint.
So what does actually get you out of a face engine?
Each engine that holds a faceprint of you has its own opt-out / erasure process, and you have to go through them one by one:
- An explicit removal / opt-out request to each engine, usually requiring you to prove which faces are yours (often by submitting a reference photo to the very database you're trying to leave). See our PimEyes removal walkthrough and the full list of databases we file against.
- Follow-up when they reject or stall. Opt-outs get bounced for blurry reference photos, partial matches, or "we couldn't confirm it's you." Re-filing correctly is half the work — see common rejection fixes.
- Ongoing monitoring. Removal is not one-and-done. Engines recrawl constantly. A new scraped photo can put you back in the index weeks after you were cleared, which is why removal has to be a standing process, not a single request.
This is exactly the work we automate: identifying which engines hold a faceprint of you, filing the right erasure request to each, handling rejections, and re-filing when you reappear.
"But it still shows up on Google" — the deindexing question
This is the other half, and we want to be straight with you about it because it's where most services quietly overpromise. If your photo still appears in a regular Google or Bing search, that's System 2 — the search index — and it is a separate problem from the face engines:
- If the source page is gone but Google still shows it, that's a stale cache. Google's own "Remove Outdated Content" tool exists for exactly this — it tells Google to recheck a dead URL and drop it. It works on pages that are actually gone (404/removed), not on live pages.
- If the source page is still live, deindexing requires getting the page itself changed or removed (asking the site owner), or qualifying under Google's "results about you" / personal-info removal policies. The image stays where it is; you're only affecting whether Google points to it.
- Cached thumbnails can linger for a while even after the source is gone, until the crawler revisits. This is frustrating but usually temporary, and it is not the same as being in a biometric database.
Why the face-engine removal is the one that matters most
It's tempting to focus on the Google result you can see, because it's visible. But the database faceprint is the more dangerous record, for a simple reason: a search-engine result requires the photo to exist somewhere public. A faceprint doesn't. The faceprint is what lets a stranger take a single candid photo of you on the street, upload it, and get your name and social profiles back — even if every photo you ever posted is deleted.
Deindexing cleans up the trail you can see. Removing your faceprint shrinks what can be done with a photo you've never even seen. If you only have the energy for one, it's the second.
Deleting the photo was the right instinct — it just doesn't reach the database that copied it.
We find which facial-recognition engines hold a faceprint of you, file the erasure requests, handle the rejections, and keep monitoring so you don't quietly reappear. That's the part deletion can't do.
Start your removals →