Google Lens will not tell you that the face on a profile also belongs to someone else’s Instagram. It wasn’t built to. It’s excellent at finding the exact same photo elsewhere online, but it deliberately skips identity matching across different photos of the same face.
That gap is why “I reverse-image-searched him and got nothing” means less than people think. You need at least two engines, run correctly, before a blank result means anything.
What each engine actually does
Google Lens (images.google.com) finds exact or near-exact copies of a specific image: the same photo, maybe resized or lightly cropped, elsewhere online. Largest index of the three, but it will not connect two different photos of the same face.
Yandex Images is built for faces. It’s the strongest general-purpose engine for facial similarity, finding look-alikes Google and Bing miss, with roughly 65 to 75 percent success in one published test. It surfaces a different photo of the same face under a different name, exactly the case you care about.
TinEye is for provenance: when and where a specific image first appeared. It sorts by oldest result, useful for proving a stolen stock or modeling photo predates the dating profile. No facial matching here either.
Rule of thumb: Lens and TinEye answer “is this exact photo stolen.” Yandex answers “does this face show up under another name.” Run at least two, and if only one, make it Yandex.
There are also purpose-built face engines, PimEyes and FaceCheck.ID, far more accurate (PimEyes 85 to 95 percent on clear frontal photos, FaceCheck 75 to 80 percent frontal, dropping to 40 to 45 percent on low light or side angles). Save these for when the free three come up empty and the stakes justify them; both carry real caveats below.
Exact steps
Getting the photo off a dating app: screenshot the profile and crop tightly to just the face, removing the app’s interface. The screenshot also strips any leftover metadata, though platforms already strip it on upload, so don’t bother hunting for EXIF data. It’s not there.
Desktop (Google Lens): go to images.google.com, click the camera icon, upload the crop, drag and drop, or paste a URL. In Chrome, right-click any image and choose “Search with Google Lens.”
Android: open the photo, tap Share, choose “Search with Google Lens,” crop to the face, then check both “Exact matches” and “Visual matches.”
iPhone: use the Google app or Chrome’s Lens icon, or open Safari and request the desktop site at images.google.com.
Yandex: go to yandex.com/images, click the camera icon, upload the crop or paste its URL. It returns a “similar” results row, where face matches live.
- Save every profile photo.
- Crop each to the face.
- Run the crop through Yandex first, then Lens, then TinEye.
- If several photos look professionally shot and all three return nothing, check the AI-generation tells below.
What a hit means, and what no hit means
A hit under a different name is the actionable signal: stop. It doesn’t matter how good the conversation has been.
No hit proves nothing by itself. Private people’s photos often return zero results because there’s nothing indexed to find. A photo stolen from someone’s private, locked Instagram also returns zero results, because the source was never public. Absence of a match is not evidence of a real, unattached person; it just means the search didn’t find one.
One useful inversion: if the photos look professionally shot and searches turn up nothing, that’s itself worth noting. Real people’s professional photos usually exist somewhere. Images with no findable origin anywhere increasingly mean they were never taken of anyone: they were generated. Check the other AI tells too: a perfectly symmetric face, blurred earrings or teeth, background text that doesn’t quite resolve, hair blending into the background, only one or two photos total.
How scammers defeat reverse search, and what to do about it
Scammers can and do bypass reverse image search by cropping, mirroring, or lightly enhancing a photo before posting it. Pixel-matching engines, which is most of what Lens and TinEye rely on, fail on a re-shot or edited image. Face-embedding engines like Yandex, PimEyes, and FaceCheck survive cropping and flipping better, though filters, low resolution, and side angles still cut face-search accuracy roughly in half. What to do about each:
- Mirrored image: flip it horizontally yourself and search that too. It sometimes surfaces a match the original crop missed.
- Cropped or filtered image: search it with Yandex specifically, since it holds up better against these changes than pixel-matching engines do.
- AI-regenerated photos: these have no origin to find, by definition. No engine returns a hit because the image was never a photo of a real person. Treat a full set of professional-looking, search-blank photos as a reason to fall back on the rest of a screening workflow: an unannounced video call, identity checks by name and city, and watching for early pressure to move off-platform or talk about money.
About PimEyes and FaceCheck.ID: legality and cost
Both are more accurate than the free three, and both carry real problems.
PimEyes’ terms of service say it’s meant for searching your own face; searching someone else’s is a breach of those terms. Five Illinois residents sued PimEyes under the state’s biometric privacy law (BIPA) in 2022 and 2023. The Cook County State’s Attorney’s September 2024 consumer alert warns that PimEyes has been used for cyberstalking and doxxing, and that its promise to block Illinois users didn’t reliably work. Free searches return only blurred thumbnails; full results run roughly $30 to $300 a month, as of September 2026.
FaceCheck.ID sells credits (roughly $19 for 150, $47 for 400, at 3 credits per search) and only accepts crypto payment, itself flagged by reviewers as a trust red flag.
Treat both as tools of last resort: only for a specific, serious reason, never re-shared, never used out of curiosity.
The bottom line
Run Yandex and Google Lens (and TinEye for a suspected stolen stock photo) on every profile photo before you share anything personal. A hit under another name ends the conversation. A blank result means “keep going with the rest of the screening,” not “verified.”
The full printable screening workflow, including this step, is the free download on this page.
Sources: reverseface.com engine comparison; socialcatfish.com reverse-image-search guide; digitalforensics.com on scammers bypassing reverse image search; Cook County State’s Attorney consumer alert on PimEyes, September 2024; Peiffer Wolf on the PimEyes BIPA lawsuits, 2022 to 2023; PimEyes premium pricing page; mobileappdaily.com FaceCheck.ID review. General education, not legal advice.