Beware your metadata trail

…at least if you’re planning to commit crimes.  The British Times Online reports that the EXIF metadata embedded in digital camera images could be used to track down whoever photographed each page of the final Harry Potter novel & uploaded it prior to the book’s release:

The information, known as Exchangeable Image File Format (Exif) data, has already revealed that the camera used was a Canon Rebel 350. Because the model is three years old, the device would likely have been serviced at least once since it was purchased, in which case the owner’s name would be known. [Via]

The reality in this case, I think, is that identifying and prosecuting the shooter would be difficult.  The camera owner would have had to have registered the camera and have had it serviced, and even with a name authorities would have to demonstrate that the person then used the camera to photograph the pages and upload the results.

Still, it’s another interesting example of digital devices recording more fingerprints than most people expect.  As devices get smarter, they’ll leave a longer trail of breadcrumbs–for better and for worse.  (How much info must be contained in an image from a GPS-enabled cellphone camera, for example?)

Adobe often ends up in a tug of war: some people really want to make metadata secure, while others want easy ways to strip it away.  Photographers seem most sensitive in this regard, wanting to ensure that their copyright info is preserved, while optionally stripping out revealing details of how, when, and where an image was captured.  On other occasions I’ve heard law enforcement folks wish aloud that Photoshop automatically inserted some trackable info into each file based on serial number.  (Don’t worry: that’s been met with an immediate, “Um yeeeah, noooo…”)

In any case, I think there’s some low-hanging fruit here.  We should offer a simple script that would let Bridge remove metadata from images, trusting that most people would use it for good & not for evil.  If you have other suggestions, please let us know.

0 thoughts on “Beware your metadata trail

  1. Yes Please!
    I’d like the option of removing the EXIF data with a script. I don’t do it often but it’s a pain going through line by line.
    [Should the script simply remove EXIF data (as opposed to other data), or should it remove only lens and camera parameters? –J.]

  2. I like the idea of me being able to lock specific data I choose into a photo that would then be impossible to remove. But how much bigger would the file have to be to tie that metadata to each and every pixel?
    Not workable I suspect.
    The best that can be done, I figure, is to make some sort of way of tagging a RAW file as unmodified. And then you can always go back to that to prove that it’s your photo.
    Courts have never accepted photographs at face value without the accompanying photographer there to testify that yes, indeed, that photo is an accurate representation of what it purports to be. So Photoshop really doesn’t change things legally to begin with.
    As far as me stripping metadata away from a photo. What’s going to stop law enforcement from using the old canard of “If you don’t have anything to hide…” Like assuming that if one takes the Fifth, or refuses to have one’s car or home searched, that means they are guilty. I suspect removing metadata doesn’t necessarily protect one’s self. Although for many situations, that will do the job just fine. But I can’t imagine now wanting all that stuff to be in ever photo I take.

  3. I prefer total editing ability — if I want it all, I add it. If I don’t want it. I have iView strip it out (too many steps for Bridge).
    I only wish I could edit EXIF, like creation date. It’s always locked, because the camera supposedly knows it’s settings best. But sometimes that internal clock is off. iView, Bridge, LR — it’s so hard to get at that EXIF.
    And sometimes you don’t need your print provider telling your client that he doesn’t like the camera the photographer (me) used: “I’D never use a Martian Camera”

  4. My first though after reading the article was that the owner can be found even if they haven’t registered their camera.
    If the owner has posted any other photos on the web with the metadata intact then in theory the photos could be searched for and found. The owner could then be identified by the context of the newly discovered posted photos. It wouldn’t be easy, but it’s far from impossible.
    Say they have an account on flickr or pbase, for example. Do a camera search, download a bunch of photos from different accounts, run some kind or script that searches within the metadata for the serial number, repeat as necessary.

  5. I’d really like to see Lightroom and Bridge allow me to specify exactly which metadata fields to preserve and which to remove when exporting images. LR currently has a “Minimize embedded metadata” option for exports, which is quite vague. Allowing me to be selective would be a big improvement.

  6. Oh no! I think you’re giving Cheney and Gonzales ideas. Now we’re going to see the Patriot Act amended to require registration of digital cameras (in a secret midnight conference committee addition to an omnibus appropriations bill that nobody will know about until months after Bush signs it).
    The Soviet Union required registration of typewriters to aid in prosecution of subversive samizdat, so why not digital cameras? We all know that terrorists use digital cameras, and copyright infringement is a form of domestic terrorism.

  7. Simon’s approach is great. If that will take too long to implement, perhaps we could have a script that strips out metadata “a la iView” in the meantime. Thanks!

  8. Well, actually Lightroom SORT OF allows for embedding of handpicked-customized metadata (rather than just stripping it, which is also an option), although it’s not implemented in a very elegant way.
    Here is how I use it: (steps 1 and 2 are only needed ONCE) 1) using one image create a set of metadata items you want (e.g., for exporting to flickr I use my flickr screenname, not my actual name; you can also add a CC license, etc.); 2) save this set as a metadata preset; 3) before you export images as jpgs: after selecting all pictures you want to export, right click and choose “make virtual copies” – the newly created vc’s will still remain selected; 4) with vc’s still selected, apply the saved (flickr) metadata set to them, and then export; 5) delete vc’s after export or, if you want to remember what you placed online where and when (as I do), save them as a collection (I have a collection folder called “Online”, with sub-collection folders for flickr, Picasa, etc. and keep those collections there.
    Be forewarned, though: Lightroom sometimes misses (i.e., fails to display and therefore modify) some EXIF fields. For example, the “user comment” that my Nikon D40 inserts (w/ my name, phone, and web address) is totally beyond Lightroom’s reach, and in fact does not even show in LR; I didn’t know it was kept on export, until I looked at the EXIF data on my uploaded photos on flickr, and there it was, even though it was NOT a part of the custom metadata set! The only way to get rid of it was to turn it off in the camera…

  9. I think tracking ownership is one thing. No. As far as preserving anything else I don’t care. The only way that works is to put my name across the picture like with Web Photo Gallery security option- makeing my name big and black enough so if somebody wants to clone it out- God bless em.

  10. When are photographers and writers going to realize they need to start pooling their energies towards innovative, offensive ideas instead of this 27/7 defensive stance of a white-knuckled, industrial age curmudgeon’s existence?
    [What are you talking about? And how come you get three extra hours in your day relative to the rest of us? –J.]
    People have to start coming to the realization that if someone wants to hack it, crack it, rip it or strip it, there is no single technological solution on planet earth that will stop them. Whatever is done, can also be undone.

  11. Hey John,
    Just thought I’d mention there already is a Bridge script to strip metadata! It’s inside the latest version of Dr. Brown’s 1-2-3.
    -Leon
    [Thanks, Leon; I’ll check it out. –J.]

  12. I’d much prefer to see digital cameras be made in Law Enforcement models which have encrypted data locked into the images. This solves (sort of) the data integrity problem most endemic to legal cases. A special version of Photoshop might well go hand-in-hand for this forensic market.
    As for metadata in Bridge, I’d actually like to see a full editor within Bridge that allows very granular control, but also ties into my personal vision of a database within Bridge. Have it respect regular expressions, and you have a serious tool.

Leave a Reply

Your email address will not be published. Required fields are marked *