FAQ

Things people actually ask.

Answers to the questions that have come up most often. If yours isn't here, the contact link in the footer goes straight to a person.

Privacy

How long do you keep my images?
Up to one hour. Both your upload and the blurred export are stored on S3 with a lifecycle rule that auto-deletes them. We never copy them anywhere else, and nothing is associated with you because there are no accounts.
Do you strip EXIF and GPS data on export?
Yes — every export. The Sharp pipeline that applies your blur regions also strips all EXIF, GPS, ICC and XMP metadata. The output file is the visible pixels and nothing else. We consider this non-negotiable for a privacy tool.
Do you use my images to train AI?
No. The third-party vision APIs we call (AWS Rekognition, Google Cloud Vision) have data-usage terms that prohibit this for their AI services. We never log image content, we never copy images into a training dataset, and we don't operate any internal ML pipeline.
Where does the upload actually go?
A private S3 bucket in your chosen region. Your browser uploads directly to S3 using a one-time presigned URL — the bytes don't pass through our application server. The bucket has 'Block Public Access' enabled and the only way to read it is a signed URL we generate when you click export.

Detection

What can you auto-detect?
Faces, license plates, document text with PII patterns (emails, phone numbers, IBANs, SSN/NI numbers, credit cards), screens, monitors, mobile phones, ID cards, and a long tail of named objects. Each category uses the strongest available API behind a single interface.
How accurate is license-plate detection?
Good in clear conditions, imperfect on oblique angles, small/distant plates, or multi-line stacked plates. We use Google Cloud Vision's Object Localization, which has a real 'Vehicle registration plate' class. If a plate is missed, draw a region by hand — the editor is built around that being a normal action, not a fallback.
Can you find the address on a letter?
We OCR document-style text with Google Cloud Vision DOCUMENT_TEXT_DETECTION, then a local regex pack flags anything matching email / phone / IBAN / SSN / Luhn-valid card-number patterns. Free-form addresses won't match a regex, so you'll often need to draw those regions manually.
Do you support video?
Not yet. Blurify is single-image for now. Video frame-by-frame editing is on the roadmap but is a substantially different product.

Technical

What's the max file size?
20 MB per image. Up to 20 images per batch. These limits keep latency predictable; if you have a use case that needs more, write in.
What image formats do you accept?
JPEG, PNG, WebP, HEIC/HEIF. We re-encode the export to JPEG by default, with WebP and PNG as options. We don't currently support animated images (GIF / animated WebP).
Does it work on mobile?
Yes — the editor uses pointer events that work for touch, mouse and pen. On phones the layout collapses to a single column and the canvas takes full width.
Can I run Blurify offline?
Not at the moment — the detection step depends on cloud vision APIs that don't ship with the browser. The blur application could in principle run locally; that's a future avenue if there's demand.

About the project

Is Blurify free?
Free to use today, with no account required. As Blurify grows we'll add paid tiers for higher volumes, batch processing, and team workflows — the core editor stays free. Paid features are about scale and convenience, never about putting basic privacy behind a paywall.
Open source?
Parts of it might end up open source. The detection adapter pattern in particular is the kind of thing that's more useful as a library than a private moat.
How can I report a bug or suggest a feature?
Send us a message — there's a contact form on the About page. We usually reply within a day.

Still curious? Read the longer how-it-works walkthrough →