Even as incident response (IR) has evolved, it has struggled to see beyond individual events and create a more complete perspective. IR tools are still very effective, particularly as advances in orchestration and automation technology have turned many IR tools into SOAR tools, but they are limited by this narrow focus.
Taking a kill chain-based perspective is the obvious next step in the evolution of IR, because serious cyber attacks are rarely single events. They are more likely to look like timelines of events that are connected in ways that are not always obvious.
In a previous column, I talked about what the cyber kill chain is, how MITRE built on that concept to create the ATT&CK matrix, and why it is valuable to security operations teams. In this article, I’d like to zoom in on incident response specifically, and look at how conceptualizing cyber attacks at the level of the kill chain is the best way to detect and disrupt serious attacks.
For a great example of how a serious cyber attack is usually not a single event, but a series of steps towards a goal, we can take a quick detour to examine a prominent recent example of cyber crime: the bank heists by the FIN7 criminal organization using the CARBANAK backdoor.
This highly lucrative series of thefts were not simply a matter of gaining access to a bank’s network and extracting funds from ATMs. They were lengthy operations that sometimes lasted for multiple months. One 2018 attack against a European bank involved spear-phishing, vulnerability scanning, domain controller compromise, Cobalt Strike Beacon, host compromise, remote access, exfiltration to command servers, and more.
Throughout all these actions, the attackers expertly kept a low profile, making detection extremely difficult. They carried out most of their activities during business hours, so as to blend in with normal activity. But certain activities, such as data exfiltration, were done in the evenings and on weekends, and limited to short sessions to avoid traffic spikes that might be noticed by bank employees.
Breaking the Chain
For reasons that should now be obvious, taking an event-based perspective to detection and response is not the ideal way to defend against an advanced persistent threat like CARBANAK. Because of the sophistication of the attackers, many of their actions may appear legitimate when viewed in isolation. It is only when viewed in the context of the kill chain of a spear-phishing attack that the malicious intent is revealed.
From a conventional incident response perspective, blocking the single action is the goal. So, one of the spear-phishing emails gets flagged and the sender gets blocked. But this doesn’t address the other elements of the ongoing attack. There may be a dozen other emails that evaded detection and you’ve done nothing to find them. With the kill chain perspective, you can recognize that a spear-phishing attack is taking place and start looking for the other links in the chain. Building off the one malicious email that was detected, you can search for other emails to likely spear-phishing targets. If one of these attempts was successful, you can look for other traces of the attack based on where the attacker is likely to go next. With this framework, a web of correlations will emerge that begins to encompass many events, indicators of compromise, endpoints, and external parties. As the kill chain is uncovered, you get the best possible chance to disrupt the chain before the attacker reaches their ultimate goal, whatever it may be.
A Step Ahead
As IR technology evolves, the term “incident response” is almost a misnomer, because it no longer needs to be a purely reactive process. IR used to serve the functions of minimizing the impact of attacks, understanding exactly what occurred, and closing the vulnerabilities that were exploited. Now, with a kill chain perspective, the “incident” to which you are responding might just be a small piece of an attack, so you have the opportunity to proactively intervene before any damage is done.
Adversaries always seem to be a step ahead, and event-based response, signature-based detection, and other traditional security methods will never completely close this gap, no matter how well they are executed. Security teams need to incorporate intent-based response, behavior-based detection, and a kill chain perspective to get a step ahead of their adversaries. The technology to support this shift is beginning to emerge, which makes me optimistic for the future of incident response.