How secure is Flickr's “disable download of original image” feature?

Question

In the Flickr privacy settings page there's a setting entitled "Who can access your original image files?". If I set this to anything other than "Anyone" (e.g. "Only you" or "Your contacts"), is it still possible for an unauthorised user to access the original image without someone giving them the URL for it? (Assume that my original photo is larger than 1024px, so there's an Original version distinct from the Large version. Also assume it's not Creative Commons licensed.)

I'm well aware that once an image has appeared in their browser, a determined user can easily download it regardless of any disincentives (e.g. JavaScript blockers) the browser tries to put in their way. However, I believe the following are correct:

1. An unauthorised user will just see an error page if they try to view the Original Size page on Flickr (e.g. this page).

2. Although the URL of that page is easily guessable (just add sizes/o/ to the end of the regular photo page URL), the URL of the actual original image file has a random component and cannot easily be guessed.

There are plenty of people on Flickr and elsewhere saying the disable download setting is useless, but I haven't seen any proof. Does anyone know for certain that it can be bypassed? If you say yes, I'll expect you to prove it by sending me the original size of my latest image! (It's meant to be available to friends & family only - so not you, Uncle Goober...)

Some context: I should point out I'm not looking to steal photos, I'm trying to understand how safe mine are, specifically with regard to this geofences loophole that was reported today.

Answer 1

I did some investigation on my own, using my own flickr account and a non-logged in browser.

Here's the All Sizes page for one of my photos.

Prior to me changing the "Who can access your original image files?" setting in Privacy & Permissions, a generic Internet user could see the "Original" link in addition to the other sizes. That page had an <img> tag that linked to this url. The "All Sizes" page also had a link that said Download the Original size of this photo. (If you check the URLS, note that there's a _d suffix on the file name; Flickr will see this and trigger the HTTP header that tells the browser to download instead of display the image).

For comparison, here's the Large size page and the corresponding image URL.

Then I changed the privacy setting, cleared the cache on my non-logged-in browser, and recheceked the links. Here's what I found:

• The link to the original size page now redirects to the large size page. That's reasonable.
• The All Sizes page didn't have the Original size links any more, as expected.
• I was still able to download the original size image
  • This is a bit surprising. It means that, while there's access restrictions on the pages containing the images, theres no security on the images themselves.
  • As a web developer I can understand why they probably did this. The images are large and static and probably served via a content delivery network. It's faster/more efficient to not check permissions for image files; you can simply host them on a "dumb" web server that way.

So, once the URL for the original file is known, there's no way to stop someone from downloading the original version of the file (short of deleting it entirely... and that may not even work. I didn't try).

One last issue: how guessable are the original file URLS? Here they are side-by-side:

Large:    http://farm7.static.flickr.com/6126/6044833128_cc02cf41e3_b.jpg
Original: http://farm7.static.flickr.com/6126/6044833128_3b8eac89d7_o.jpg

So, the suffix (_b or _o) determines the size, but there's also another element in the filename that varies depending on the size. You can't just change the suffix to flip sizes. Here's the URL for the Large version with the suffix switched to _o; it doesn't work.

If I was Flickr, I'd make sure that that middle element was completely random per photo size, and hence unguessable except by brute force attack. It's 40 bits long, so there's a lot (2^40, ~1 Trillion) of possible options. It's very unlikely anyone would bother to brute-force that segment just to get the original size version of a file... when they already have the large version.

So, as long as you've turned off the "Original file download" feature and you don't share the URLs of the original images, I'd say that the Flickr feature is pretty secure. If it breaks, it's pretty much your own fault.

Answer 2

This page (linked to from the Firefox plugin that bill weaver posted) does a good job of summarising the situation, including the "random component" in the image URL that I mentioned in the question.

The author notes:

This means that even if you go to the trouble of getting the file name for one of the smaller sizes, you cannot guess the file name of the original photo, and this is great news for photographers worried about image theft.

"Cannot guess" - great! :) But then he goes on to say:

The cool thing is that after Flickr randomized the file names, it became next to impossible to guess the URL for a file’s original size.

"Next to impossible" - not so great! :( But I assume he just means that given enough time and processing power you could crack it with a brute-force attack. If so, that's good enough for me: I'll take those odds. :)

Answer 3

Flickr uses non-secure web protocol (HTTP) by default, so any images can be accessed after performing session hijacking from a person who can access them. For session hijacking, the attacker needs to be able to eavesdrop on victim's network traffic, e.g. by accessing same wireless access point or some intermediate network node. The risk has become quite significant in public wireless spots after Firesheep was released - a Firefox plugin that automatically grabs cookies other people use in the same wireless area and presents them for easy use.

Also, the images come with headers that allow intermediate web caches to keep them for years. So getting access to browse a web cache might also provide access to original images that have been viewed through that cache.

I am not going to chase down your or your friends' network connection or caches, so no, I am not going to send you your original image. But to play safe, your best bet is not to upload original images.

Answer 4

If it's displayed, of course you can save it. If originals are protected, then not unless the URL is known. Bottom line is pretty secure and no i don't think it can be downloaded.

The flickr-original firefox plugin seems at first to do this (i passively assumed so), but actually downloads the large size if you've protected your originals from public viewing. I've downloaded a 1024 x 580 version of your photo with this plugin.

Answer 5

Firefox plugin Tamperdata will allow the discovery of the protected url for the image. It is trivial to do so. The only layer of security appears to be obscurity.

Answer 6

Here's the easy way to do this (using Safari) without all the jargon you folks are throwing out there. Go to the all images page. Once you've enabled your Develop drop down menu go to Show Web Inspector. On the left side of the developer window, look for the images drop down. Click that and look for the string with the long number. When you click that, the image will appear in the developer window. On the far right you will see the image URL. Copy and paste the URL in a new browser window and the image appears and can be downloaded from there. OR just drag the image from the developer window. (It will likely have its name changed to Unknown.