Richard Bejtlich’s NDR Guide Advocates Evidence-Led Hunting and Network Interdiction

Summary
Richard Bejtlich’s NDR guide recommends using network evidence to test threat-hunting hypotheses, disrupt malicious activity before attackers reach their goals, and apply AI with human verification.
Key points
- Bejtlich argues that alerts should begin an investigation, not define its conclusion; analysts should gather evidence to test or disprove a hypothesis.
- The guide identifies full packet captures, extracted files, transaction logs, and alerts or detections as sources of network evidence.
- Network interdiction aims to identify and disrupt malicious activity after an initial compromise but before attackers achieve their objectives.
- Suggested hunting leads include unusual protocols, large outbound transfers, lateral movement, executable identification, and certificate exposure.
- AI can support triage, playbook execution, and coordination across tools, but the article emphasizes human verification to limit hallucinations and unintended consequences.
- The guide recommends a “zero-baseline” approach to reduce alert fatigue from pre-enabled rules.
Article Details
- Defense Focus
- Use network detection and response to validate incidents with defensible network evidence and disrupt malicious activity before attackers achieve their objectives.
- Detection Methods
- Start threat hunts with a hypothesis about adversarial techniques, then query network logs and sessions to validate or disprove it.
- Identify executables and investigate unusual protocols.
- Track large outbound data transfers and detect lateral movement.
- Analyze certificate exposure.
- Use alerts as the beginning of an investigation, not as conclusive evidence of an event.
- Data Sources
- Full packet captures
- Extracted files
- Transaction logs
- Alerts and detections
- Network logs and sessions
- Rule Types
- Alert rules; the article recommends a zero-baseline strategy to reduce alert fatigue.
- Defensive Actions
- Isolate and contain malicious activity after initial compromise and before a full breach.
- Collect and validate network evidence to support or reject an investigative hypothesis.
- Use network visibility to identify and disrupt malicious activity before attackers achieve their objectives.
- Require human verification of AI-assisted findings and govern automation to prevent hallucinations or unintended consequences.
