Slashdot Mirror
Nikon's Image Authentication Insecure
silanea writes "Elcomsoft claims to have broken Nikon's Image Authentication system which — apparently only in theory — ensures that a photograph is authentic and not tampered with through a digital signature. They were able to extract the signing key from a camera and use it to have a modified image pass the software verification, rendering the rather expensive feature mostly marketed to law enforcement all but useless. So far Nikon has not given a statement. Canon's competing system was cracked by the same company last December."
Whew - I've always hated having to wear a ski mask when I "work". Now I can just claim image tampering.
He's getting rather old, but he's a good mouse.
This is great news, because now people will be able to cast doubt on images when there is cause to instead of being told "it's not possible it's a fake, it's signed". You know that if someone cracked it publicly someone else (probably many someone else's) have cracked it in private, and have kept around the ability to forge photographs in case of emergency... that ability is now reduced.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Just take a picture of the photo-shopped image with your Nikon camera. Bam! That was sure hard to crack.
Has there ever been a case whose outcome depended on the authenticity of a digital image?
If I remember correclty, three or four years ago a driver in the UK got out of a fine because he was able to prove that the photo used as evidence was faked. I don't remember the details, I think he parked in a car park and they tried to claim he overstayed using a doctored photograph as evidence?
I think the authorities will still say "it's not possible it's a fake, it's signed" and it'll be up to the victim (or the victim's lawyer) to know that the signage has been broken.
The last time I was stopped in a speed trap (on motorcycle), I knew it was coming up (they always put a speed trap in this particular construction zone on weekends because people ignore the temporary "35" signs 'cause there's nobody working on Sunday, but I digress) and had slowed way down before taking the turn, but was waved over anyway. I was pretty sure he'd tracked the (obviously faster) car one lane over instead of me, and said so. He said "the gun can't be wrong, I had a firm lock on you." I can see the stupid radar gun in his hand right there, and it's not like there's a scope on it, or even if he actually had me in crosshairs, that it could tell the difference between a slow moving object in the foreground and a much faster object in the background. I maintained that he could not possibly have locked on me, because he would have read 33 MPH, which is what my speedo was displaying at the time. I said it obviously had "locked" on the car that passed me shortly after the corner. The cop said that this was impossible, radar guns don't make that kind of mistake.
Well hell, there's a huge body of evidence that radar guns make "mistakes" all the time. I laid out exactly how the error could have occurred, he continued to insist that the gun can't make mistakes. I finally said "ok, whatever. We'll see what the judge says." He went away, talked to his cohorts for awhile, came back and issued me a "verbal warning", let me go. Now, I strongly suspect that if I'd acted like I knew nothing about the technical details of radar guns, I'd have gotten a ticket.
Oliver's law of assumed responsibility: If you're seen fixing it, you will be blamed for breaking it.
As long as the signing key is unique per-camera(which I would bloody well hope it is, for forensic purposes), "tamper-evident" is arguably good enough, and probably easier to approach(as with any hardware security measure, the approach to the ideal is more or less asymptotic, with price spiking to near infinity as you reach the goal).
If the camera is tamper-evident, anybody who suspects manipulation of photos ostensibly from that camera can attack the credibility of the camera on technical grounds, just as they might a witness: "Your honor, the camera has probe traces on its 'secure' ROM pins, its private key could easily have signed more shit than John Hancock."
In such a case, anybody who wanted to use the camera for evidence gathering would be required to maintain physical security around it, as is necessary.
The problem crops up if the key can be extracted silently, or is shared between multiple cameras. Tamper-proof is optional. Tamper-evident is absolutely necessary, or doubt is cast on every image signed.
Yeah, just look at that thriving Xbox 360 pirate game environment.
With enough effort any DRM can be broken somehow. The only thing the content owner has to do is ensure that it's difficult and/or expensive enough to not be worth it.
And in case you didn't read the..summary, the camera is supposed to help provide a chain of evidence. Not sure why anyone would put "feature" in double quotes and act like nobody would buy a camera that supports this.
I didn't know that. Can you point me to any information about this? I'm googling here and not coming up with anything about emulsion negatives containing holograms, but probably because I'm not formulating the search very well.
If you have any links I'd really appreciate it.
You are welcome on my lawn.
"Kodachrome" is exactly right. I worked in a large (3M) film processing lab (factory, really) back in the early 70's. Probably once a week, the local sheriff or PD would have an officer come by with some SLIDE (chrome) FILM (typically autopsy or crime-scene photos) to process - they'd stand by and watch while their film was processed - maintaining the CHAIN OF CUSTODY at all times, and requiring signatures from workers when the film was out of sight (like in a darkroom). They never had their slides mounted, they'd just walk away with the whole processed roll. I don't know for a fact, but I'd be willing to bet that it's still the standard way to handle film evidence (that really matters) even today - except on NCIS or CSI.
No folly is more costly than the folly of intolerant idealism. - Winston Churchill
If you think of the emulsion layer as being a three-dimensional object that has some depth to it, instead of a two-dimensional plane with zero thickness, I believe that you'll find that it is obvious: It will not be exposed equally throughout that depth, and there will be definite and observable paths that the light has followed within the emulsion layer.
I don't know if I'd call it "holographic," just due to the confusion that the term itself presents in common use (as GP pointed out), but it seems like an adequate and correct description nonetheless.
Kid-proof tablet..
And also prove you don't have a second camera that has been tampered with to have the same key as the untampered camera?
'It is unique per camera, it says so in the press release which I linked in a separate post.'
I may be missing something, but I can't see this in the press release, so there may well just be a single key. However, every camera model with the image authentication feature also writes its (unencrypted) serial number to an EXIF tag. If image authentication had remained secure, you could have 'proved' which camera took the photo simply by reading the serial number from the metadata of an authenticated image (tampering with the number would invalidate the image).