Threat Intelligence teams often rely on the Common Vulnerability Scoring System (CVSS) to triage, assess, and escalate vulnerabilities impacting their technology stacks. However, there are sometimes delays in assigning a CVSS score by industry authorities, such as the National Vulnerability Database (NVD), which is maintained by the National Institute of Standards and Technology (NIST).
As the volume of CVEs has continued to increase year over year, the NVD has struggled to keep up with the increased demand for analysis and scoring. The following report, captured on June 18, 2025, from NIST, showcases the vulnerabilities awaiting analysis.
As a result, organizations face challenges including:
In response to these challenges, Feedly developed a CVSS estimate score to bridge the gap between CVE identification and CVSS score assignment, providing threat intelligence teams with an early warning system to monitor, assess, and prioritize CVEs with higher fidelity.
Feedly uses a multifaceted approach built on machine learning models to estimate CVSS scores. The estimate is calculated based on the vulnerability type, description, attack complexity, and exploit info.
Feedly continues to update its CVSS Estimate until the NVD assigns an official score; thereafter, it no longer updates the CVSS estimate. Security teams should use Feedly’s CVSS estimate as an interim measure and switch to the official score once available.
When an article appears in Feedly, it is assessed by a series of machine learning models to analyze and predict CVSS based on the attack vector, vulnerability description, and past CVEs with confirmed CVSS scores. Knowing these past attack vectors and their relationship with CVSS scoring based on years of collected data.
Feedly deploys a system of machine learning models trained on historical data where the inputs are articles mentioning the CVE. The predicted outputs represent the NVD CVSS categories which are used to calculate CVSS scores. Feedly will assess and score weighted vocabulary, historical datapoints, and patterns across articles mentioning specific CVEs to arrive at aggregated CVSS estimate scores. Feedly's CVSS Estimate system contains the following structured process:
Feedly AI builds and maintains a comprehensive dataset derived from thousands of past CVE reports. Each data point is scored based on its “importance” in relation to CVSS scoring. These associations enable machine learning models to rapidly identify vulnerabilities that are more likely to be severe, based on the data points collected in CVE and trusted OSINT reporting.
Feedly AI then predicts the probability of a specified outcome, in this case, the likelihood that a CVE has a High (7.0-10.0), Medium (4.0-6.9), or Low (0.1-3.9) CVSS score. It assesses against historical data, determining which patterns of data are indicative of a respective score. This machine learning engine calculates the probability that a CVE belongs to one of these categories based on the weighting of these words ingested from source inputs.
Lastly, Feedly AI assesses combinations of data points collected and aggregates the predictions to determine a final CVSS Estimate. This machine learning model can capture complex patterns in data to arrive at a final estimated score.
Feedly’s CVSS Estimate appears in both CVE Insights Cards and the Vulnerability Dashboard, providing threat intelligence teams with a way to better track and monitor vulnerabilities impacting their technology stacks.
Enable the CVSS Estimate column in your Vulnerability Dashboard to view Feedly’s estimated CVSS score for each CVE. As shown in the image below, even when an official score hasn’t been published yet, Feedly provides an estimated CVSS to help you assess severity early.
This feature is especially useful when scanning multiple CVEs at once and applying the dashboard filters to highlight only the most relevant vulnerabilities and their estimated impact.
When deep-diving into a specific CVE, the CVE Timeline in Feedly's Insights Cards offers valuable context, including the CVSS Estimate Score. In the example below, Feedly flagged the CVSS as HIGH even before NVD officially assigned it a score of 8.8.
CVE Insights Cards are free and publicly available. Even if you're not a Feedly customer, you can explore them at feedly.com/cve to get a full 360° view of any CVE you're tracking.
In sectors where early warning systems are crucial for effective threat response, threat intelligence teams can bridge the gap between official CVSS Score assignments using Feedly’s CVSS Estimate capability, moving towards near-real-time assessment of vulnerability-related threats. By combining CVE data enrichment, machine learning models, and data-driven severity predictions, Feedly’s CVSS Estimate enables teams to efficiently identify, assess, and escalate critical vulnerabilities, facilitating faster and informed decision-making to protect their organization’s critical technology stack.
If you would like to explore CVSS Estimate scores further, please contact Feedly’s Threat Intelligence Advisory team via Enterprise support at enterprise@feedly.com or start a free trial.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。