EPSS (Exploit Prediction Scoring System)
You'll learn how machine learning predicts which vulnerabilities are most likely to be exploited in the wild.
After scanning systems and collecting CVSS scores (which tell you how bad a vulnerability is), you face a critical question: which of these thousands of vulnerabilities should you patch first? CVSS severity alone doesn't tell you which flaws attackers are actually targeting. This is where EPSS comes in.
What EPSS Does
EPSS uses machine learning models trained on historical exploitation data to produce a probability score (0-100%) that a specific CVE will be exploited in the next 30 days. Instead of just severity, it considers factors like:
- Whether exploit code exists publicly
- Social media chatter and threat intelligence mentions
- Vulnerability age and disclosure trends
- Historical exploitation patterns
- Vendor and product popularity
Think of CVSS as describing the height of a cliff (how dangerous a fall would be), while EPSS estimates the likelihood someone will actually fall off that cliff.
Practical Application
A CVE might have a CVSS score of 9.8 (critical) but an EPSS score of only 2%—meaning it's severe but rarely exploited. Conversely, a moderate CVSS 6.5 vulnerability with a 75% EPSS score is actively being weaponized. Your remediation priority should favor the latter.
EPSS scores update daily as new intelligence arrives. You combine CVSS (impact) with EPSS (probability) and your environmental factors (exposure, asset criticality) to build a risk-informed patching queue that reflects real-world threat actor behavior, not just theoretical severity.
Key Takeaway:
EPSS uses machine learning to predict exploitation probability, helping you prioritize patching based on actual attacker behavior rather than severity alone.