Security analysts are a CISO’s most valuable sec ops resource. Plain and simple.
And with cybersecurity expertise in high demand and low supply, retention of those security analysts is a top priority among businesses everywhere. Not to mention, the sec ops skills gap will only widen in 2018. A recent survey revealed that more than half of all respondents have “a problematic shortage of cybersecurity skills.”
In response, many businesses are eying human-augmented machine learning for continuous threat monitoring. On the surface, security automation assumes many of the tedious responsibilities of the security analyst, which helps smaller teams accomplish more in less time.
But on a deeper level, eliminating the “noise” associated with traditional SIEM solutions opens up security analysts’ schedules to more judgement and experience-based functions that boost employee engagement while improving organizational security posture.
Despite the name “human-augmented AI,” it’s hardly just the AI that’s being augmented; it’s also the depth of capability of the human analysts who underpin your security operations center.
As of mid-2017, 79 percent of security analysts said that they were overwhelmed by the volume of threat alerts. Simultaneously, nearly, three quarters of those same respondents felt that they were also being stretched too thin with vulnerability maintenance tasks.
We certainly need to acknowledge the obvious here: An overwhelmed sec ops team is an ineffective sec ops team. Without time and resources for adequate alert triage, misses that result in harm to an organization are far more likely to occur.
But once again, there’s a longer-term consideration at play. The average salary of a security analyst is $96,000 a year, and the current unemployment rate among these professionals is microscopically low, according to U.S. Bureau of Labor Statistics. Meanwhile, there are an estimated 28,400 vacancies, according to U.S. News & World Report -- and that’s the conservative estimate. Others estimate unfilled cybersecurity jobs to be in the millions.
The point is, alert fatigue isn’t just a liability for your data security. It’s also a detriment to the very life-force of your entire SOC: your analysts. If they aren’t fulfilled, they will leave, and replacing them will not be easy, to say the least.
The point of security automation is not to replace humans. The simple truth is that ML is not developed enough to monitor, rate, triage and hunt for threats, let alone respond to them, with total autonomy.
However, current security automation platforms function by architecting a continuous feedback loop by which analysts can train ML in a much more nuanced way than simply programming it with threat intelligence about “safe” or “good” events from the outset. Therefore, analysts actually begin to curate AI's responses to, and understanding of, events with middle-of-the-road threat ratings.
Under this model, clear threats, such as malware with documented signatures, will be dealt with automatically. Likewise, recurring false positives will be given low threat ratings and sifted out, and will also be used to deeply contextualize future false positives so they never actually reach analysts for investigation.
Only the most questionable activities and behaviors will be escalated to the analyst, who will then be responsible for performing thoughtful evaluations rather than rushing through endless lines of alerts only to realize they’ve been chasing their own tails.
Give the people what they want: thoughtful, judgement-based roles.
Leave the rest to the machines.