When it comes to Internet threats, the correct response to the Shakespearean question, “What’s in a name?” ought to be “Who cares?” according to Mike Banic.
“The important thing is to look at what a threat is doing, not what it is,” he told an audience at SOURCE Boston 2016 this week, in a talk titled, “Understanding Attackers’ Use of Covert Communications.”
“There seems to be a lot of pride in naming threats,” he said, “but a lot of them behave in similar ways, and you don’t need a signature to recognize that. The IP address and the URL may change, but the fundamental behavior will not.”
Banic, vice president of marketing at Vectra Networks, one of about three dozen presenters at the annual event, said given the reality that “the perimeter is really porous,” effective security means being able to detect when an attacker is on the inside.
And these days, he said, signature-based defenses are so last generation. Attackers generally will not use an IP address or domain name more than once, he said, “so if you’re collecting logs, that’s like studying for the test by looking at the paper of the guy who flunked.”
Once attackers have successfully dropped malware into a network, he said, they have numerous ways of remaining hidden. “They can control both ends of communication,” he said. “Any app, protocol and encryption is available. Hidden communication is the underlying enabler of modern attacks. If attackers are going to steal, they are going to be low and slow.”
They can hide within encrypted traffic, which more and more frequently is encrypted by default. Or they can “hide in plain sight” by using the “draft” folders of standard email applications.
“They (messages) are auto-saved to the server and the cloud,” he said. They never need to be sent. And the person who controls both endpoints doesn’t even need to encrypt. An attacker can read it on the outside. Think about how hard it would be for a firewall to pick that up.
“You should expect obfuscation – hiding is the name of the game,” he said.
But, Banic said, attackers inevitably create behavior patterns that can be detected through the use of machine learning algorithms. He invoked the declaration of the iconic investigator, Sherlock Holmes: “While the individual man is an insoluble puzzle, in the aggregate he becomes mathematical certainty.”
The same principle applies to malicious intruders, he said. “In the aggregate, it becomes easy to find them. Traffic analysis can find threats without decryption.”
There is a difference, he said, between “opportunistic” attacks, which cast a wide net and use compromised devices to attack others, and “targeted” attacks, which are what the name suggests.
In targeted attacks, he said, eventually, “somebody puts their fingers to the keyboard,” and begins to exhibit behavior that can be tracked and analyzed. “Malicious behaviors are similar across platforms,” he said. “You’re looking for flow patterns at the packet level. It’s a whole sequence that causes the algorithm to fire.”
He cited the example of a remote-access Trojan (RAT) called GlassRAT that went undetected for several years until RSA discovered and reported on it late last year.
It targeted Chinese nationals associated with large, multinational corporations.
It escaped detection by antivirus tools, “and was highly successful at avoiding signatures,” Banic said, “but when we investigated it, the algorithm fired within 15 minutes.”