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.

