Threat hunters emerge from the shadows

As a new generation of cyber threats emerges, a new breed of cyber security defenders is emerging. They don’t wait for enemies to strike, but go looking for them, writes ARTHUR GOLDSTUCK.

Meet Alex and Andy. They are young, clean-cut and casually dressed men who may not draw a second glance if you bump into them in public. Behind closed doors, however, they are the information age’s equivalent of fearsome warriors. They are called threat hunters.

No, they are not super-heroes. Although they are heroes to some organisations.

Five years ago, the job description didn’t even exist. Back then, security experts waited for the hackers to strike, then rushed around repairing the damage, fixing the holes and, just maybe, chasing down the bad guys.

But that’s not good enough today. As hackers become more sophisticated in both their methods and their tools, and the stakes get higher in lost data and massive financial fraud, the security industry also has to evolve.

Alex and Andy represent the cutting edge of this evolution. In an age of heavy reliance on algorithms and artificial intelligence to predict and block standard threats, it turns out that human intuition is far more powerful in spotting unusual and new kinds of attack.

They work in Johannesburg for a global cyber security consultancy called MWR InfoSecurity, which develops tailored security solutiosn for clients that range from governments to corporations.

Their boss, Jacques Louw, MWR director and head of Cyber Defence, describes threat hunting as a technique that “focuses on the human elements in attack detection”. In this environment, he says, “one cannot have automated threat hunting”.

He uses the evolution of physical home security as an analogy for the growing need for a new kind of approach.

“We’ve seen in the last decade or so that, ultimately, you cannot prevent attackers from climbing over your walls no matter how high you build them; they always seem to find a higher ladder. So the focus has really been towards detection and response for when they do manage to get over. In this regard, detection is critical – if you don’t see the intruder, then the armed response never arrives, whereas if you have too many false alarms, the armed response will stop showing up.”

Add the fact that organised crime syndicates are now targeting major corporations, and state sponsored teams are going after national governments, the challenge becomes even more complex. Enter a new kind of detection.

“Detecting attacks is not a new idea,” says Louw. “In fact it’s been around for many years, comprising automated systems like anti-virus and intrusion detection or prevention systems. Think of them as anti-virus for a network traffic.

“Such systems were originally built to deal with viruses or malicious software or malware that was self-propagating – which means it spreads on its own in an automated way. After the malware is created, it acts automatically, executing the tasks it was initially programmed to do.

“Traditional anti-virus works by analysing a piece of malware, creating a signature for it and adding it to a database – similar to that of a book of criminal records. The anti-virus then checks each new program on a system against these records, and alerts you when it finds one with features similar to that of one of the records.

“Unfortunately, the first problem here is that you need to discover a piece of malware before you can create a signature for it, so it has to succeed in attacking someone first before a signature can exist. Moreover, it is quite easy for a programmer to change what the malware looks like – while still having the malware do the same thing. So changing features, but not behaviour.”

A key reason traditional security is no longer enough is that the threat has evolved from the equivalent of a property invasion to something far more elaborate.

Louw compares a modern corporate network to a large city, with multiple roads in and out and many thousands of everyday people performing many different activities across the city.

“In this analogy, a signature-based system is similar to a simplistic robot that walks around the city, trying to match faces of people he sees to a photo in the mugshot book. In contrast, threat hunting is like a human policeman that can actually spot bad behaviour, not because each bad activity has been strictly defined and given to him on a long list, but rather because he can use his experience, knowledge of the law and judgement to make a call on some behaviour that has never been seen before.”

An example is a form of malware that operates not like a virus spreading through the system, but like a hole in the system. Appropriately called RATS, for Remote Access Trojans, these are “exploits” that can easily be altered slightly to bypass anti-virus every time they are used.

“The attacker only needs to find a single entry-point to compromise an organisation, so the defender must defend all systems perfectly all the time to be secure,” says Louw. This concept is known in security as “the defender’s dilemma”

“In threat hunting, one of the ways we are responding to this dilemma is by using a judo-like technique called the anomaly analysis, where we can turn the attackers greatest advantage into a disadvantage. Instead of looking for something bad on a large network of systems, we look for something that is different and investigate it accordingly.

“If the attacker attacks a system on the network, that system will appear different to all of the other systems in some way – allowing the attack to be detected even if we don’t have a signature for exactly what the attacker is doing.  This may sound simple, but you can easily recognise that certain differences are more interesting than others and that is where human skill comes into play.

“You can have automated systems gathering data from all systems, collecting network traffic and pulling in logs from systems. Ultimately, however, you need a human to drive the analysis.”