PS-4 · Threat Intelligence & AI

Honeypot-to-Threat-Intel Pipeline Using Hybrid ML/LLM Architecture

Focus

Threat Intelligence & AI

Objectives

  • Turn raw honeypot observations into structured threat-intelligence records.
  • Combine ML classification with language-model summarisation and validate both.

Deliverables

  • A pipeline from raw logs to structured intelligence records
  • An evaluation of the ML and LLM stages separately

Constraints

  • Honeypot data must be anonymised before it leaves the lab.
  • Language-model outputs must be checked against source records.

Expected outcomes

  • Structured, verifiable threat-intelligence output.

Safe-testing boundary

Use synthetic or pre-anonymised honeypot logs.