/

Anomaly Detection

Anomaly Detection is active by default in every Monitor project. It learns your application's normal behavior and automatically flags activity that deviates from it, detecting unknown threats, abnormal patterns, and suspicious access.


What It Does

Anomaly Detection builds a behavioral baseline from your application's real usage and continuously analyzes it to identify suspicious activity.

It monitors:

  • Authentication patterns: failed login spikes, credential rotation, login attempts from unusual locations or at unusual times
  • Session behavior: geographic jumps within a session, concurrent sessions from different locations, impossible travel scenarios
  • Device behavior: sudden changes in device usage patterns, interaction styles, and feature access sequences
  • Request behavior: abnormal request rates, non-human navigation sequences, automated patterns
  • Error patterns: sudden spikes in errors that may indicate scanning or fuzzing

Why It's Always On

Anomaly Detection doesn't require configuration because it doesn't rely on predefined rules. It builds its baseline automatically from your application's real usage and flags deviations.

This means it can detect:

  • Zero-day attacks that no signature exists for yet
  • Credential stuffing campaigns using leaked credential databases
  • Brute force attempts against authentication endpoints
  • Account takeover patterns where attackers test stolen credentials
  • Reconnaissance activity before a targeted attack
  • Automated access from bots or scripts mimicking user behavior

How It Differs From Other Protections

Unlike other ByteHide Monitor protections that require explicit configuration via BHMMonitor.configure() and a BHMProtectionModuleType, Anomaly Detection:

  • Has no enum value: It is not part of BHMProtectionModuleType and cannot be added via enableProtection()
  • Requires no setup: Works out-of-the-box on every Monitor instance
  • Learns automatically: Establishes and continuously adapts its behavioral baseline
  • Runs transparently: Operates in the background without impacting application performance

What Gets Reported

When Anomaly Detection identifies suspicious behavior, it creates an incident in your Cloud Panel with:

  • The type of anomaly detected (authentication, session, device, rate, etc.)
  • Confidence score based on how far the behavior deviates from baseline
  • Device info, session details, and contextual metadata
  • Timeline of the suspicious activity

You can review these incidents alongside incidents from other protection modules in the same dashboard.


Configuration

Anomaly Detection works out of the box. You can adjust its sensitivity through the Cloud Panel or JSON configuration:

JSON
{
  "anomalyDetection": {
    "sensitivity": "medium"
  }
}
SettingOptionsDefaultDescription
sensitivitylow, medium, highmediumHow aggressively deviations are flagged

Sensitivity Levels

LevelBehaviorBest For
LowOnly extreme deviations trigger incidentsHigh-traffic apps where minor variations are normal
MediumBalanced detection with few false positivesMost applications (default)
HighFlags subtle anomalies, more incidents to reviewSecurity-critical applications (finance, healthcare)

Previous
Overlay Detection