Attribution: Fingerprints vs. Ballistics and Inductive vs. Deductive Reasoning
In the present day where the word “Cyberwar” is all the rage, and governments as well as private sector entities are seeking to cash in on the power grab that is mostly information warfare as the Chinese actually call it (信息战) too many are forgetting a core problem to the picture. This problem, is “attribution” as it has been termed in the community. To attribute an attack to an individual, government body, or group, is something that to date, has not been discussed as much as I would like to see with regards to all of the cyberwarfare talk as well as any other inferences with regard to forensics and geopolitical ascription to acts of “war” as this is has been labeled by this terrible terminology that we have latched onto.
Nomenclature aside, there are issues around trying to determine definitively where an attack has really come from because of the nature of computer systems, varying countries that they reside in, and the potential for the actor to be anyone from nation state to individuals of a collective privately, or a single determined individual. It is my contention that “attribution” can be very hard to prove in a court of law, never mind that a country may in fact be ready to wage war against another on the grounds of what is taken to be the truth of where an attack originated from and who the actors really were. There are too many variables that may never be one hundred percent certain to be basing any of these decisions on in my view, unless one has hacked back into the core final system that originated everything and that is rarely the case today.
So, where does this leave us? How do we even attempt to attribute an attack to any one person, government, or group? Can we ever be certain of any of this information? Can we base an aggressive action against a nation based on any of it?
Fingerprints and Ballistics
Some would approach the problem of attribution of digital attacks on the methodology that began the criminal forensics process we have today. Fingerprints were the first forensic model for determining who really may have created a crime if the evidence did not consist of an eye witness attesting to the fact that “they did it” Ballistics soon followed once guns began to have lans and grooves bored into the barrels to allow for more accuracy. Both of these examples leave telltale marks on the bullets or objects to determine which person or what gun were the arbiter of whatever crime was committed. Today though, we do not have the same narrow confines of data to examine as both of these examples allow for.
Code is the medium of today and while there are certain ways to tell if code was written in the style of a person or written on a particular computer, for the most part, these do not allow for absolute certitude as to who the actor was that created the code, nor for that matter, who used said code to effect an outcome (i.e. attacks on systems) conclusively. All one really has in most cases, are pieces of code, that, with the right coder, may in fact look like anothers, or, all attributions have been stripped from, or, lastly, copied directly from open sources and then tweaked. All of these scenarios allow for a great lassitude on determination conclusively on source or origin.
With all that said, the digital fingerprints are there, and with luck someone can determine if the coder was sloppy and forgot something. Interestingly, much of this was out in the open and talked about with regard to the Stuxnet infections in Iran. Once the code was audited, there were many subtle clues as to who “may” have written, and in fact there were potential red herrings left in the code such as “mytrus” and other tidbits that may in fact just been placed there to mess with those seeking to perform forensics in hopes of finding out who did it. To date, many think that the US and the UK did the work, planned the operation, created the code, and implemented it, but, there is no conclusive proof of any of that is there?
Suffice to say, that everyone does make mistakes, but, with the right amount of diligence, it an adversary can make it incredibly hard code wise, to determine who did the writing. On the other side of the coin, the digital forensics arena also looks at the network and hardware side of the equation as well. Many attacks today are not directly coming from the home systems of the adversary, but instead they are coming from proxy machines that have either been rented or, more likely, hacked previously. This too can be heavily obfuscated and be something of a problem to gather information from if those systems reside in countries unfriendly to the attacked parties. One would likely have to hack into those already compromised systems and then attempt to gather intelligence as to where they were being controlled from and by. This is of course if the system wasn’t already burned or, as in many cases, the logging had all been removed and thus there were no logs to see.
From this perspective, yet again, there is a great amount of doubt that can be injected into the picture of just who attacked because of the nature of the technologies. Unless the systems are live, and in fact the adversary is either still using them or was exceedingly sloppy, it could be very hard to in fact prove conclusively any one actor or actors carried out and attack even from the digital forensics side of the house. This leaves us with a problem that we have to solve I think in order to truly be able to “attribute” an attack even tentatively to anyone. One cannot only rely on the technologies that are the medium of the attack, one must also use reasoning, psychology, and logic as well as whatever the forensics can allude to as to the attacker. This is very much akin to the process used by CIA analysts today and should be the SOP for anyone in this field, because the field is now truly global as well as has been brought into the nation state arena of espionage and terrorism, never mind actual warfare.
Inductive vs. Deductive Reasoning
First off, I would like to address Inductive and Deductive reasoning in this effort as one of the precepts core to these attribution attempts. By using both of these in a rigorous manner, we can attempt to shake out the truths to situations that may in fact seem clear on the face of them, but, once looked into further may be discounted, or at the very least questioned. Much of this lately has been the hue and cry that APT (Advanced Persistent Threat’s) are all pretty much originating from China. While many attacks have in fact been attributed to China, the evidence has not always been plainly clear nor, in many cases, has the evidence been anywhere in the open due to classification by the government and military.
There are many “secret squirrels” out there and they all pretty much squeek “CHINA” all the time. Unhappily, or perhaps unfortunately, these same squirrels end up being the ones talking to the news media, and thus a juggernaut is born in the news cycle. It just so happens that there are many other nation states as well as other actors (private/corporate/individual) that may well be the culprits in many of the attacks we have seen over the years as well. Unfortunately, all too many times though, a flawed inductive or deductive process of determination has been employed by those seeking to lay the blame for attacks like ghostnet or ghost rat etc. Such flawed thought processes can be shown by examples like the following;
All of the swans we have seen are white, thus, All swans are white.
This has pretty much been the mindset in the public and other areas where attacks in the recent past have been concerned. The attacks on Google for instance were alleged to have come from China, no proof was ever really given publicly to back this up, but, since the media and Google said so, well, they came from China then.. Right? While the attack may have in fact come from China, there has been no solid evidence provided, but people are willing to make inductive leaps that this is indeed the truth of it and are willing to do so on other occasions where China may have had something to gain but proof is still lacking. The same can be said with the use of deductive reasoning as well. We can deduce from circumstances that something has happened and where it may have originated (re: hacking) but, without using both the inductive method as well as the deductive with evidence to back this up, you end up just putting yourselves in the cave with the elephant trunk.
Psychology and Victimology
Another part of the picture that I believe should be added to the investigative process on attacks such as these, is the use of psychology. By using the precepts of psychological profiling as well as victimology, one can take a peek into the motivations of the attacker as well as the stance of the victim that they attacked into account on the overall picture. It is important to know the victim, their habits, their nature, and background. These factors can often lead to insights into who the adversary may in fact be. While the victimology paints the picture of the victim, it also helps flesh out the motives and possible psychology of the aggressor as well.
Of course one need not be a board certified psychiatrist or psychologist to perform a vicimtology in the way that we need to within the confines of determining who may have hacked a client. Many pentester’s do this very thing (though perhaps not enough today it seems) by profiling their targets when they are preparing for a test scenario. The good ones also not only look at what the target does, but also how they do it. They also look at how things work logically, as well as every other aspect of the business to determine how best to attack and what would have the most effect to replicate what an attacker “could” do to them. This is a key also to determining who may have actually attacked as well as why they did and this leads to another part of the puzzle, that of motives.
In trying to determine who attacked one must look at the motives for the attack. These motives can also show you the lengths that the attacker was willing to take (i.e. creating custom code and other APT style attack vectors/methods) to effect their end state goal. If there seems to be no real reason for their attack, and they have not stated it in other ways (like Anonymous and their declarations of attacks) then we are left to come to grips with seeking the reasons as well as what they took/destroyed/manipulated in the end. It is important to look at the whole picture instead of focusing on the minutiae that we in the INFOSEC field often find ourselves looking at daily in these IR events.
Hannibal Lecter: First principles, Clarice. Simplicity. Read Marcus Aurelius. Of each particular thing ask: what is it in itself? What is its nature? What does he do, this man you seek?
The Pitfalls of Attribution Theory
Another part of the picture that must also be assessed is that of the mindset of the assessor themselves. Today we seem to have quite the echo chamber going on with the likes of Beitlich and others concerning China and APT activities as I alluded to earlier. The media of course has amplified this problem threefold, but, the core problem is that we as investigators are sometimes easily tainted by the echo chamber. Thus I put it to you that the precept of “Attribution Theory” also play a key role in your assessments and that it can be a pitfall for you. In Attribution theory, one must also take into account such things as the motivations of the person doing the attributing. This means that even if you are a consultant in an IR, you too can allow your own leanings to sway your findings in such an endeavor as trying to determine who hacked whom with leading evidence but no definitive proof thereof.
Motives are key, motives of the assessor, motives of the victim, and motives of the adversary. One must take these all into account and be as impartial as possible and mindful of these things. It is my contention today, that all too often people are all too available to the idea that “China did it” is the go to assessment of a so called “APT” attack, especially so when APT is one of the most misused acronyms today in the information security field. It is just behind the term “Cyberwar” in my opinion in fact as one of the most misused and poorly constructed acronyms or terms for what is happening today.
In the end, one must take a step back and see the bigger picture as well as the minutiae that comprises its total while not being too easily swayed by our own bias or conditioning. I suggest you acquaint yourselves with these ideas and use them when involved in such cases where APT and Cyberwar are concerned.
There will Always Be “Reasonable Doubt”
In conclusion, I would like to assert that there will always be reasonable doubt in these cases. Given now that we are considering actions of war and legislation over attacks and counter attacks within the digital sphere, I would hope that those in government be made aware of the issues around attribution. I cannot conceive of going to war or launching missiles over a digital attack on some system somewhere. The only way I can see this actually becoming kinetic is if the attack is in tandem with boots on the ground or missiles fired from a distinct area of a foreign power. Unfortunately though, it seems of late, that governments are considering such actions as hacking the grid, as an acceptable trigger to kinetic response by the military. This for me is all the more scary given what I know about attribution and how hard it is in the digital world to determine who did what and when, never mind from where.
Presently I am working on a framework of this whole process model and will in the near future be presenting it as well as other aspects of determining the attribution of attacks on companies and systems at a conference in Ireland. It is my belief, with my partners in this presentation, that given more subtle cues of psychology, as well as sociological and historical inference, one can get a greater picture of the attacker as well as the motives for an attack if they are not openly stated by the aggressor. Of course none of this will eliminate “reasonable doubt” but, as CIA and other intelligence analysts have proven with such methodologies, one can make a more solid case by looking at all aspects surrounding a person, case, or incident to determine the truth.