Organizations everywhere are fighting one of the great unsolved issues of cybersecurity: insider threats. According to a report by CA Technologies, 90 percent of organizations feel vulnerable to insider threats, and 53 percent reported experiencing insider incidents in the past 12 months. Although security teams are aware of the problem, they often fall short of mitigating it because they lack the means to adequately monitor user behavior.
What Is an Insider Threat?
Before we address how organizations can enhance their user behavior analysis capabilities, let’s take a step back and explore the three main types of insider threats:
- Ignorant users who are not aware that their credentials or devices have been compromised.
- Negligent users who bypass security controls and simply don’t care what data they access or where they store it.
- Malicious insiders seeking to commit fraud or identity theft for personal gain or to deliberately harm the organization.
What do all these actors have in common? They put the organization at risk by exposing critical data to vulnerable devices, such as personal USB drives, and accessing privileges administrator accounts from unsecure locations. The challenge is to determine what type of behavior typically precedes such unauthorized access and to distinguish that behavior from legitimate activity.
4 Steps to Mature User Behavior Analytics
That’s where user behavior analytics (UBA) and machine learning come in. But Rome wasn’t built in a day: A successful UBA deployment requires careful planning and a thorough understanding of the organization’s user population, IT environment and security posture. Below are four key steps to guide security teams in their UBA implementation.
Step 1: Decide How and Where to Review User Anomalies
As important as the output of analytics is the decision on how and where to investigate any anomalies that are detected. A traditional security operations team may not be trained in all the business applications. That’s why organizations often move user behavior analytics to a dedicated internal application operated by a threat management team that is authorized and trained to investigate the depth of the threat.
Step 2: Start Collecting Access and Authentication Data
Most security teams do not have real-time or historical insights into who is accessing the enterprise environment. This should be a priority. Security teams can reveal valuable insights by configuring machine learning algorithms to ingest log data from Active Directory, Lightweight Directory Access Protocol (LDAP) or other identity and access management (IAM) tools. This should uncover threat alerts such as usage of dormant accounts, access from unusual locations and unexpected access to critical servers.
Step 3: Expand With Network and Session Data
The network never lies. Adding network data such as user proxy logs, virtual private networks (VPNs), and intrusion detection and prevention systems (IDPSs) can help surface potentially suspicious user behavior such as login anomalies, persistent Secure Shell (SSH) sessions, unusual time or geography changes, and sharing of credentials. At this stage, machine learning will have been running for multiple weeks, so the UBA solution should be able to detect abnormal behaviors such as unauthorized use of networked resources and access.
Step 4: Add Endpoint, Cloud and Application Data
This stage is where UBA really starts to shine. By adding endpoint activity logs, security teams can match identity and access anomalies to accounts. This enables organizations to explore advanced use cases, such as abnormal access into service accounts or restricted application areas. Once the baselining and initial learning stages are complete, security teams can start building peer groups and comparing individual user activity.
The Path to UBA Maturity
While these steps appear simple on the surface, security professionals must fine-tune their analysis by selecting individual user activity categories and extractions at each step of the process via a crawl-walk-run method. Output for the UBA monitoring should come in various forms, such as full user lists and associated risk scores, alerts on user activity, and session summary views that show the exact time and date when a sudden user risk change occurred.
Insider threats exist in many forms, all of which put the organization at risk. UBA can help security teams uncover access abuse and high-risk activity, but it requires a concerted effort to mature machine learning and data analysis capabilities across the organization.