INTRODUCTION

Since the standard of Common Vulnerabilities and Exposures (CVE) was first introduced in 1999, almost 200,000 publicly known vulnerabilities have been recorded to date. While many of these vulnerabilities have since been patched (some were patched years, even decades ago), many organizations have not yet applied the available security updates and patches, leaving their systems exposed to cyberattack.

Given the increasing number of vulnerabilities discovered and disclosed each year, and the mounting struggles of IT departments as they work to keep legacy systems secure, many organizations do not prioritize patch deployment for ‘low-severity’ CVEs, focusing instead on remediating those that are making headlines or those assigned a ‘high-severity’ Critical Vulnerability Scoring System (CVSS) rating.[1]

This prioritization process is inherently flawed: the CVSS score measures the estimated severity – not risk – of exploitation. In fact, some of the most widespread and devastating cyberattacks over recent years have originated with the exploitation of vulnerabilities rated ‘medium’ or ‘low’ severity by the CVSS. If anything, threat actors understand that these ‘medium’ and ‘low’ severity CVEs are likely to remain unpatched and unremediated within enterprise environments, and take advantage of the flawed prioritization process to gain access to critical assets, moving laterally through the network to deploy high-profit, high-impact attacks.

Predicating the prioritization of patching cycles on CVSS ratings alone is therefore a potentially fatal error, and yet it remains the prevalent methodology for many organizations and vulnerability scanning tools. In many cases, timely patch management for CVEs rated at medium or high severity is an issue of compliance, required by industry standards, government agencies or other regulatory bodies such as the Payment Card Industry Data Security Standard (PCI DSS).

HOW A CVSS SCORE IS CALCULATED

CVSS scoring mechanisms have gone through three major revisions (and a number of minor revisions) since the framework was inaugurated in 2005, with CVSS Version 3.1 being the most current revision. According to the National Vulnerability Database (NVD), CVSS Version 3.1 is generated through the measurement of three core metric groups: 

(1) Base Score Metrics, which represent the intrinsic and fundamental characteristics of a vulnerability; 

(2) Temporal Score Metrics, which represent the current state of exploit techniques or code availability; 

(3) Environmental Score Metrics which represent those characteristics of a vulnerability that are relevant and unique according to each individual organizational infrastructure.[2]

Published CVSS scores are typically comprised of a combination of Base Metrics and Temporal Metrics Scores only. While a useful starting point, Base Metrics by definition are static, representing the intrinsic characteristics of a vulnerability that remain  constant over time. Once it is nce published, the score is often not re-evaluated or updated to reflect the current Temporal status.Static scores are not much help in the current, rapidly evolving threat landscape. How can an organization effectively prioritize their CVE patching cycles if they rely on an outdated severity rating that remains unchanged even after a working exploit kit has been widely distributed within the cybercriminal underground?  As explicitly noted in the CVSS version 3.1 user guide, CVSS measures severity, not risk. Accordingly, insight into threat actor discourse and interest surrounding CVEs and their related attack vectors for exploitation is critical, providing the accuracy, relevance and context needed to effectively prioritize vulnerability remediation processes. 

ENRICHING CVSS

The topic of zero-day exploits and exposed vulnerabilities is always trending within cybercriminal communities, both on clear web platforms and on the underground. From 280-character tweets circulated among cybercriminals on Twitter, POC exploits released on clear web code repositories, to exploit kits and tools shared across the forums and markets of the deep and dark web, threat actor discourse revealing which vulnerabilities they plan to target is far from scarce.

Let’s examine CVE-2022-22954, a VMware Workspace ONE Access and Identity Manager vulnerability that allows remote code execution through server-side template injection.

The vulnerability currently has a CVSS 3.1 score of 9.8 – likely flagging it as the highest priority patch for organizations using the software. Though remediation of this CVE is likely to be prioritized quickly, it is important to understand the context of underground threat actor discourse surrounding the exploitation of this vulnerability.

Cybersixgill offers vulnerability contextualization in a number of ways: 

First, derived from AI analysis of underground discourse, Cybersixgill’s Dynamic Vulnerability Exploit (DVE) scoring system predicts the likelihood of vulnerability exploitation over the next 90 days, assessing the immediate risks of each vulnerability based on threat actors’ intent. The DVE score for this vulnerability is currently a 10 (scale 1 – 10), indicating that this CVE is at critical risk of exploitation. 

Secondly, each Cybersixgill DVE rating is backed by an audit trail explaining the rationale behind the score, providing visibility into all the chatter and unique attributes that support the prediction of exploitation within the next 90 days. For CVE-2022-22954, most of the chatter was observed on social media platforms such as Twitter. However, there are also a significant number of code repository entries of POC exploit codes.

It is also important to note chatter on high-profile underground forums and the extent of actor participation in those discussions. For CVE-2022-22954, chatter surrounding the exploitation of the vulnerability was observed on a notable underground forum as early as 4/12/2022.

RE-PRIORITIZING

Monitoring underground chatter does more than simply justifying what is already set to be prioritized by one’s vulnerability scanner. There are likely multiple vulnerabilities with a CVSS score below 4.0 which would go unflagged by scanners due to their low severity score. Fortunately, Cybersixgill provides intelligence even for those CVE with low CVSS severity ratings that are likely to be overlooked by security teams, yet present significant potential risks to the organization.

For example, on a well-known Telegram group dedicated to the discussion of hacking tools and tactics, a message was observed sharing information surrounding the recently publicized CVE-2022-22950 and CVE-2022-22948 vulnerabilities. It’s easy to argue that the discussion of these CVEs on such a prominent cybercriminal channel might encourage threat actors to target these vulnerabilities, regardless of their low-severity CVSS ratings.

Additional chatter was also observed on other cybercriminal Telegram groups associated with notable hacking groups discussing the same  CVEs. Monitoring this communication in real-time can support the patching cycle process, empowering security teams to prioritize remediation according to accurate threat intelligence regarding the likelihood of a vulnerability exploitation.

CONCLUSION

Vulnerability and exploit chatter is rife across all spectrums of the internet. Yet this intel can be extremely difficult to track without real-time visibility into the primary arena of cybercriminal activity – the deep and dark web – rendering an accurate identification of immediate threats a near insurmountable challenge. Cybersixgill’s advanced collection mechanisms gain and maintain access to the most extensive range of sources from the deep, dark and clear web, automatically extracting, processing and analyzing intel as it surfaces, to provide  the context needed to justify a patch of a critical vulnerability.

While all vulnerabilities ought to be of some concern, only 6% of CVEs are actually exploited. Without accurate threat intel to provide insight into the risk – rather than severity – of each vulnerability, security teams find themselves fighting  an uphill battle, overwhelmed with the sheer volume of vulnerabilities potentially exposing their organization. Automated threat intelligence helps to separate the wheat from the chaff, by providing the much-needed context to drive informed security decisions, and by helping teams enhance the productivity and efficacy of their patching cycles, without exposing their systems to avoidable risk.

 

[1] The CVSS scoring of vulnerabilities is calculated according to several variables that examine the process and potential impact of exploitation, resulting in a final severity score of low (0.0-3.9), medium (4.0 -6.9), or high (7.0 – 10.0) severity. These scores vary slightly depending on the version of CVSS applied (CVSS 2.0 or 3.1).

[2] https://nvd.nist.gov/vuln-metrics/cvss/v3-calculator