[ADDITION] exifree

[arnon-shub]

Category

Metadata Removal

Software to be added

exifree

Supporting Material

• Main site: https://exifree.com
• GitHub repo: https://github.com/arnon-shub/exifree
• Strips EXIF, GPS, device, date and all metadata from photos entirely in the browser — zero server upload
• Supports JPEG, PNG, and HEIC (auto-converts to JPEG)
• No account, no tracking, no cookies, no third-party scripts (except lazy-loaded HEIC converter from CDN when needed)
• Open source (MIT)
• Threat model / landing page explanation is on the main site

Affiliation Disclosure

I am the developer and maintainer of exifree.

Would you like to submit a PR?

None

Please tick the boxes

• You have filled out this form accurately, and to the best of your knowledge
• You have indicated whether or not you are associated with the suggested addition
• A similar submission has not already been opened for this software / service
• You agree to the code of conduct

[lissy93]

It's a neat app!

Right now, there's only desktop apps listed in the metadata removal category, which can be a barrier for some users, so there's definetly a need for something more accessible and browser based.

But I think for now, let's revisit once the project is more mature. Just because if there are any issues, it usually takes a few weeks/months of real-world use before they get found/reported.

[arnon-shub]

Makes total sense, happy to wait. Just pushed a security fix based on your suggestions in the other issue. Ping me whenever you want to revisit.

[minus21grams]

Pretty handy app privacy-wise, but it increases the file size significantly.
Ran some tests using some old, 12 MP (3072 x 4096), photos. Here's the results:
Original photo: 2.768 KB ➜ Cleaned: 3.299 KB
Original photo: 2.881 KB ➜ Cleaned: 3.448 KB
Original photo: 2.833 KB ➜ Cleaned: 3.381 KB

[ltguillaume]

Indeed, this isn't a lossless operation, it just reads the image file and saves it as a new image (with a compression setting that results in larger files in most cases).

If you're particular about keeping the original image and quality, I highly recommend stripping the EXIF info some other, lossless, way, because this is just transcoding.

[lissy93]

Yes, and right now it only works for a very narrow happy-path, specifically just JPEG.
That could be okay, except everything else gives an illusion of working, but actually does nothing. Which is worse than just erroring, since it gives a false sense of security.

Metadata removal is hard. Each file format has its own structure and encoding, and the metadata can live in multiple layers and locations throughout that structure, and sometimes be so intertwined with the image/file content that it's just impossible to remove (looking at you AVIF/HEIF!).

So OP is a braver person than me by attempting it!! If it were me, I'd forget the idea of building it from scratch, and use something proven, like libvips or ImageMagick (in WASM, so it's still browser-based and offline). And that'd also make the shortcomings possible to document more transparently, because it's still never gonna be 100%

[arnon-shub]

Thanks for the detailed feedback — it was genuinely useful.

Here's what we changed based on your comments:

1. Lossless binary stripping (no more Canvas re-encoding)
   The previous approach re-encoded through Canvas, which changed file size and wasn't truly lossless. It's now replaced with direct binary stripping:

• JPEG: removes APP1–APP15 segments (EXIF, XMP, ICC, IPTC, Adobe) and COM blocks. Image scan data is untouched.
• PNG: removes tEXt, iTXt, zTXt, eXIf, tIME, iCCP, pHYs, cHRM, gAMA, sRGB chunks. IDAT is never touched.
• WebP: removes EXIF, XMP, ICCP chunks and clears the corresponding flags in the VP8X header.

Output file size is now nearly identical to the original.

2. PNG and WebP actually read and strip metadata
   Previously, non-JPEG files showed "No metadata found" even when metadata existed — giving a false sense of security. Now the parsers read the correct chunk structures for each format.

3. Self-hosted HEIC converter
   The heic-to library is now bundled locally instead of loaded from jsDelivr CDN. Zero external requests.

4. Honest disclaimers
   The "after stripping" panel now states that this is binary-level removal, not re-encoding. Edge cases in exotic formats (e.g. deeply embedded AVIF metadata) are acknowledged.

Edge cases in exotic formats (e.g. deeply embedded AVIF metadata) are acknowledged in the UI.

[arnon-shub]

A few more updates since the last comment:

Batch processing — drop multiple photos at once. Each is stripped silently; a single ZIP is generated with a summary: "3 photos · saved 1.2 MB".

Size before/after — single-file mode now shows original vs. clean file size inline, making the lossless nature of binary stripping visible.

Accessibility — scored 100% on Safari's built-in accessibility audit (22 checks passed, 0 warnings).

[lissy93]

That looks really good now!! 💪

Just a quick one - try downloading and then testing with this image (it's safe, but shows your original xss is still present).
And some images are getting corruped after stripping, for example this pic.

Sorry, but I'm not smart enough to understand where in the code the corruption issue gets introduced 😅😪

[arnon-shub]

Both issues fixed:

XSS — metadata values are now inserted via textContent instead of innerHTML, so injected markup is displayed as plain text and never executed.

Image corruption — the JPEG stripper was pushing bytes one-by-one into an array, which caused issues on larger files. Rewrote it using Uint8Array slices. Tested with both files you shared — both strip cleanly now.

Thanks for catching these.