There's a common fallacy in popular culture that depicts robots as an improved, mechanical version of the human form. The Tin Man from The Wizard of Oz, Gort from The Day the Earth Stood Still, Rosie from The Jetsons,and C-3PO from Star Wars are just a few classic examples of that mistaken assumption. Actual humanish robots don't do as well as their science fiction counterparts, as this compilation clip from the DARPA Robotics Competition gloriously illustrates.
Humans are all-purpose survivors. Robots, on the other hand, are at their best when designed to carry out specific tasks. Don't think Tin Man; think ZX200S. Don't think Rosie; think Roomba. Robots can accomplish the things they are designed to do faster and more accurately than people — and without fatigue or complaint. The fatigue part is critical. At our best, humans are prone to make mistakes. And while practice improves our performance, fatigue will quickly degrade both our motor functions and our decision-making. To protect public safety, truck drivers and pilots are limited by law from operating their vehicles for long periods without adequate rest. And with evidence that driving drowsy can be as dangerous as driving intoxicated, many states have laws against fatigued driving by private motorists.
Fatigue is also an issue in cybersecurity — one that can lead to poor decision-making. On their own, security analysts can't be expected to keep pace with the onslaught of threats. The volume and persistence of attacks against the enterprise are simply too high and too constant. By some calculations, networks may be subject to thousands of attempts to breach security every day, and if yours is a high-priority target, that number is far greater. It is estimated that the U.S. Department of Defense is bombarded by a withering 36 million daily email-borne attacks. Even at our best, human beings can't handle those kinds of numbers. We need help.
Fortunately, we are seeing greater use of automation and machine-driven decision-making tools to combat this problem. As always, new categories of cybersecurity technology have emerged to meet growing threats and are giving enterprises a means of not only thwarting the adversary, but of overcoming the pitfalls that come with relying solely on human intervention.
Here are some predominant ways automation is transforming cybersecurity:
• Practice and Repetition for Preparedness
Security's biggest challenge was trying to prove a negative. No breach must mean that the technology and defense is working. Yet at some point, compromise is confirmed, and sometimes it existed for an extended period — and rarely is anyone ready. The reality is that true preparedness — and from it, we hope, prevention — comes from two models that must be understood and repeated, as often if possible, and with automated technology assistance where available. The two viewpoints that need to be driven into organizational muscle memory are "paths, not point-of-attack" and "plan execution, not point-in-time."
It's crucial to understand that one compromise does not make a breach. Finding the point of entry is only the beginning of the "kill chain," and organizations need to find as many of the subsequent paths to the ultimate target — and then also track the potential paths back out. Automation is huge in calculating the universe of paths and probabilities. Once this is done, repetition — both in practicing attacks and coordinating response — are the critical complement. Industry analyst firm Gartner has outlined some of the technologies in these two categories, terming them breach and attack simulation and security orchestration and response, respectively.
• Correlation of Data for Incident Response
One of the biggest challenges for security analysts when they receive an alert is understanding whether the threat is real or not (that is, a false positive). A variety of data from different sources is needed to make that decision. For example, if a malware alert is triggered, an analyst might want to pull the logs for the device associated with the IP address in question, validate the owner and role of the device with a network access control system, and check the malware sample in VirusTotal (to see if another security product has detected this variant). Security automation and orchestration platforms can simplify this process, enabling multiple sources of information to be pulled together into one screen for analysts to query, eliminating one of cybersecurity's most time-consuming operational tasks.
•Security Analytics for Threat Detection
The first phase of threat detection technology that emerged gave security teams the ability to detonate files with potential infection in a virtual sandbox — the advanced persistent threat or zero-day malware solution. Once a file was found to be infected, a signature for the malware in question was created, extending protections beyond "patient zero." New automation and analytics solutions expand beyond this capability to include processing and analyzing massive amounts of user and network data to identify anomalies. Automation enables faster and more accurate analysis and, combined with machine learning, brings greater insights to potential threats lurking in the environment.
Automation in cybersecurity brings the promise of scalability of resources, maximization of efficiencies, and increased effectiveness. What is critical is to ensure that vendors offering each of the promising technologies above do not further bog down security teams with more "noise."
Tomorrow's security operations centers won't be staffed by robotic bipeds with lasers for eyes and synthesized voices asking the hackers they detect to "take me to your leader." But they will be — and indeed are — outfitted by sophisticated systems that use machine learning, intelligent automation, and indefatigable vigilance to keep hackers from succeeding at their craft. I, for one, welcome our robotic security overlords.