Anonymization is three jobs, not one
The obvious job is the pixels: faces, name tags, license plates, a screen with an email address on it. The second job is the file itself — phone photos carry EXIF metadata that usually includes GPS coordinates precise enough to find the doorstep where the picture was taken. The third is context: a caption, a distinctive uniform, a recognizable storefront. A tool can handle the first two. The third one is a thirty-second read of what you're about to post.
Hide every face — and choose how
Drop the photo into the editor and every face is detected and outlined automatically — AWS Rekognition draws a tight box per face, and crowd shots come back with each face found separately. You decide which ones to hide: toggle any box off to keep a face, or draw an extra region if someone's half-turned head slipped past the detector.
The style matters more than people think. A gaussian blur reads as soft and works for most social posts. Pixelation is more deliberate — the mosaic look makes it obvious something was removed, which is the convention in journalism. Solid fill is the strongest option: it leaves nothing of the original region to reconstruct. Whichever you pick, cover the whole face, hairline to chin — if you can still guess the expression, go stronger.
Every style is a destructive pixel edit in the export. The covered region's original pixels are replaced, not overlaid, so no amount of editing the downloaded file brings them back.
Strip what the file knows
EXIF metadata is the part of anonymization most people forget. Beyond GPS coordinates, the file can carry the exact capture time and the device model — enough to place a person somewhere at a specific moment. Blurify strips EXIF and GPS from every export automatically. If metadata is your only concern and the pixels are fine as they are, that works too: export without blurring anything and the file still comes out clean.
How EXIF removal worksWhat GDPR actually asks of a photo
Under GDPR, a photo of a recognizable person is personal data, and publishing it generally needs a lawful basis — often consent. Where consent isn't practical, blurring and pixelation are widely used to reduce identifiability, and stripping location metadata removes another identifying layer.
Two honest caveats. Anonymization is judged by outcome, not by tool: if clothing, location or the caption still identify the person, the image isn't anonymous, however thorough the blur. And none of this is legal advice — if you're processing photos at scale or in a regulated context, that's a conversation with a professional, not a blog post.
What the tool side can promise: Blurify never stores face descriptors, embeddings or any biometric data. Detection boxes live in your browser tab, and the image itself is deleted from storage within an hour.
A quick checklist before you post
Every face covered, hairline to chin, including reflections and the person half-visible at the frame's edge. Name tags, plates, screens and paperwork covered. Metadata stripped — automatic if the file came out of Blurify. Caption and surroundings read once with fresh eyes: would someone who knows the person or the place recognize it anyway? If any answer is shaky, fix it before it ships — a photo can be re-posted forever, but it only needs to leak once.