Verified Forensic Image

Hi Samr,
I would say a forensic engineer with a potentially false impression that his tools are functioning correctly is 1000% worse than a noob.(because the engineer has the ability to get far deeper into the material)

Hopfully this is a forum for Professionals and we cannot help out the new guys if all we do is lower the bar and do not do the best we can to fully inform them of the requirments and instill in then a sense of responsability and professionalism.

Yes maybe the guy is 'playing' at forensics as a hobby but that is a decision he has to make not us.

With that in mind I always take the view to offer the best information I can with cited references from reliable sources, so that they can go dig the material up.

I don't want to be the one who's comments are web searched ,then incorrectly appled that allows that pedophile to get away. (I was recently speaking to a detective who told me they were so overworked & undertrained they were in somecases searching the forensic websites for information)
Now If they do not have the time to fully read the thread, who knows where it will end.

P.S I disagree on the MD5, there have been a number of cases where MD5 has been called into question.
On the issue of the MD5 being 'more' secure on larger files, it is actually the opposite.
Any hashing algorithm is a 'loss of information', the more information involved in calculating the MD5 hash, the more space you have to 'play about'. ALL Hashing algorithms have collisions, the issue with MD5 & SHA1 is that you can back work the case to alter the information in the original file and still maintain the original hash.

It ia all about 'reasonable doubt' and a exposing non-technical people(judge jury) to a highly technical subject of cryptographic processing.

So you hash with as many systems as possible (MD5/SHA2, etc), and if you get 'called out on MD5' you roll out the big guns.

C.

Hi Beetle,

I had actually done a long reply, but the system appears to have swallowed it.

I was acutually talking about "reasonable doubt", If i can show the MD5 system is flawed I can call it into question when presenting evidence in a court.

As in this case

http//blogs.zdnet.com/Ou/?p=409&tag=nl.e589

"The motorist's defense lawyer took advantage of the courts ignorance and argued that the MD5 hashing algorithm was a discredited piece of technology and therefore the speeding photos were invalid. "

P.S I disagree on the MD5, there have been a number of cases where MD5 has been called into question.
On the issue of the MD5 being 'more' secure on larger files, it is actually the opposite.
Any hashing algorithm is a 'loss of information', the more information involved in calculating the MD5 hash, the more space you have to 'play about'. ALL Hashing algorithms have collisions, the issue with MD5 & SHA1 is that you can back work the case to alter the information in the original file and still maintain the original hash.

It ia all about 'reasonable doubt' and a exposing non-technical people(judge jury) to a highly technical subject of cryptographic processing.

So you hash with as many systems as possible (MD5/SHA2, etc), and if you get 'called out on MD5' you roll out the big guns.

C.

Although there are mathematical attacks on MD5 in respect of crafting a second pre-image that would generate the same digest output, it becomes a matter of whether the existence of such attacks would cause the purpose of the generated hash to be questioned. If the hash is being used to tie extracted data to the original data as it existed on a physical device seized from the accused it is much the same as the lands and grooves in ballistics. You are tying two objects together to show the integrity of the extraction (tying the bullet from the victim to the gun that you put in the accused's hands through fingerprints). You are proving the integrity of the extracted information in relation to the source, not providing authentication services, which is where the real concerns regarding MD5 and SHA1 actually lie. How likely is it that the data extracted has the same hash as another file on the media seized? When trying to 'prove' the validity of an image, tying it to a physical device, keeping in mind the huge (in a bit-wise sense) variances in the layout of drives and manufacturing defects, how likely are you to find two drives in the possession of the accused that would generate the same hash? I submit the answer to both is it would be astronomically small - similar to two unrelated persons, say you and I, in the same room having the same DNA.

Hashes are not the be all and end all of proof, there is also the overall context of the data in question, as well as other testimony or real evidence that puts the data into the prosecution's theory of the crime.

As a side note

Reasonable doubt is generally defined as reasonable doubt in context of the totality of all the evidence, not a fanciful or theoretical issue (I heard one judge a few years ago referring to not "falling prey to flights of fancy"). What you really are referring to in reasonable doubt are findings of guilt (beyond a reasonable doubt). If the court has concerns about the accuracy of evidence it will affect the weight attached to that evidence during deliberation. I can't recall in 36+ years in LE the term 'reasonable doubt' being attached to evidence itself. If the evidence of a hash is called into question it generally will require expert testimony to assist the court into assigning weight.

Hi Beetle,

I had actually done a long reply, but the system appears to have swallowed it.

I was acutually talking about "reasonable doubt", If i can show the MD5 system is flawed I can call it into question when presenting evidence in a court.

As in this case

http//blogs.zdnet.com/Ou/?p=409&tag=nl.e589

"The motorist's defense lawyer took advantage of the courts ignorance and argued that the MD5 hashing algorithm was a discredited piece of technology and therefore the speeding photos were invalid. "

This didn't demonstrate that the MD5 system is flawed for this particular application, it just demonstrates that the opposing counsel didn't properly counter the defense lawyer's claims.

-David

Hi Beetle,

I had actually done a long reply, but the system appears to have swallowed it.

I was acutually talking about "reasonable doubt", If i can show the MD5 system is flawed I can call it into question when presenting evidence in a court.

As in this case

http//blogs.zdnet.com/Ou/?p=409&tag=nl.e589

"The motorist's defense lawyer took advantage of the courts ignorance and argued that the MD5 hashing algorithm was a discredited piece of technology and therefore the speeding photos were invalid. "

This didn't demonstrate that the MD5 system is flawed for this particular application, it just demonstrates that the opposing counsel didn't properly counter the defense lawyer's claims.

-David

Absolutely agree with you David.

I would like to read the transcript of this one as it sounds like the defence
a) tendered evidence -and-
b) gave opinion evidence which is a form of hearsay (R v Abadom).

Both huge no-nos.

Hi Kovar,

The link was not to highlight a case where the MD5 hash had been fabricated.
Merely a link showing that due to the research a lawyer was rightly or wrongly able to get the case thrown out. There is a link to the court transcript and the later clarification by a secondary judge, about the lack of an expert witness.

The issue still remains that if an image or a file can be show to have content that can vary and still have the same hash, then such systems are not suitable methods for validation (for whatever purpose).
Irrelevant of the fact that the changes do or don't impact the evidence.

Hi Kovar,

There is a link to the court transcript and the later clarification by a secondary judge, about the lack of an expert witness.

I see the link to the newspaper article (MD5 Defence), but not the one to the transcript…
Where is it?

RB

code_slave,

(interesting handle by the way)

To even show that hash collisions are relevant you would have to demonstrate/explain what they are, how they can occur and the likelihood of collisions in the context of the case. Better yet take some of the existing evidence, create a collision for it and then provide the court with a statistical model of how it could happen at random or some proof of the accused's expertise to create a collision.

I am not sure what you mean by the evidence impact comment. I think that the spectre of potentially fabricated evidence would have a significant impact.

I think this thread has gone beyond the question in the originating post…

I would offer the opinion that there is two general types of testing for hardware and software, official and unofficial. Official begins at NIST and includes the arena of verification and proof of accuracy in criminal and civil court. Unofficial being for the examiners own satisfaction and curiosity including his professional obligation to understand his tools. A good examiner will have an insatiable curiosity about everything and will be doing it for his own intellectual satisfaction without even considering that he is "supposed to" anyway. I believe that Kaly falls under this category.

You will generally find that most forensic hardware and software has already been certified as admissible or not in most courtrooms. Because of this the unofficial testing becomes more significant to individual examiners. This opinion is not absolute and is subject to change for any reason at any time.