Many security conversations today are likely to touch upon security intelligence, behavioral anomaly detection and the value of augmented intelligence (AI) in security. But is cognitive security all hype, or are there real applications in use today? Is it feasible only for the largest, most sophisticated organizations, or is it more widely available?
We’ve heard about AI’s usefulness in healthcare diagnostics, its ability to interpret and filter massive amounts of data on clinical trials for overworked physicians who can’t keep pace with newly produced studies, and even its ability to propose tailored treatment strategies based on an individual’s DNA traits. But cybersecurity is different. It has traditionally been a continuous process of incorporating common access controls at known vulnerable locations based on rules and policies. Controls are then monitored for success and to identify areas requiring additional rules and policy enhancements. These are fundamental security practices that are applied effectively every day across almost every organization using similar tools, practices and skill sets. It has been the accepted logical approach for so long — why change now?
Most organizations will tell you that they have a mature security framework in place and that their practice is the most effective model. Their challenges usually include inadequate resources such as skills, tools, people, funding and time to keep up with the steadily increasing demand. But does their current framework truly scale with the evolving threat landscape and resourcing trends? Is it reactive or proactive? Many organizations around the world are now facing similar challenges as the healthcare industry — challenges that can be met with the help of cognitive security.
What Is Cognitive Security?
Before we continue, let’s distinguish true cognitive security from basic behavioral anomaly detection. Behavioral anomalies are really just basic table stakes for any AI solution today. You may not even require true AI to detect anomalies because patterns, rules and polices can surface such alerts.
True cognitive security, on the other hand, is interpretive based on continuous learning that increases its comprehensive corpus of knowledge. While it certainly does identify behavioral inconsistencies, it also goes much further. A cognitive solution can conduct its own assessment of the subject at hand and develop its own hypothesis, freeing security analysts from the task of defining strict rules and traps. It can provide insight that would otherwise be elusive, and do it considerably faster than humans. In fact, AI might be more correctly referred to as “accelerated intelligence.”
Streamlining Threat Assessment With AI
Let’s review the work a security team typically performs on a daily basis. An analyst may notice a suspicious event and begin an initial assessment to determine whether to investigate the anomaly further or ignore it. Their initial conclusion may be based on the team’s interpretation of the situation, their skills and the tools available to it. The team might also examine an event notification from its security controls to determine whether it’s a false positive or a true alert. Again, it’s typically up to the initial analyst to decide whether to act on or ignore an event.
During this initial decision-making process, analysts have myriad tools, channels and forums at their disposal. An analyst might try to summarize his or her suspicion in conversations with others, conduct inquiries within technical or security forums, initiate keyword searches on traditional public search engines, or look through internal repositories to review incident response playbooks.
Consider how much time is exhausted during initial triage before any remediation begins. A cognitive security solution can conduct all of this research simultaneously, examine more sources of information than a team of analysts would have available to them and generate conclusions within minutes. It can also determine whether an event is real, its threat origin, remediation information, payload pathways and other subtle indications of a possible attack. AI can even evaluate payloads traversing throughout the infrastructure and identify other users who may have received the same payload. This information presents very quickly, reducing the time it takes to detect and respond to incidents from days to minutes.
Accelerating the Future of Security
While cognitive computing technology is still in its infancy, these tools are delivering real value to many organizations today and growing faster and more intelligent by the minute. The threat landscape is expanding at an unprecedented pace, and bad actors, many of whom are heavily funded and well-orchestrated, are also embracing the newest technologies, including AI.
Many sophisticated threat actors are investing heavily in research and development and purchasing illicit products anonymously with digital currencies. As a result, these malicious wares are developed and deployed more quickly, growing more advanced and becoming resilient to traditional security measures. Given this rapid development, it seems we’re headed for a future in which AI-enabled security tools must interpret and respond to AI-powered threats — if we’re not there already.