# Guided Lab: Identifying a Poisoned Model Artifact

> A hands-on guided lab demonstrating how to detect hidden backdoors in AI models using activation clustering and adversarial probing.

Canonical: https://codelivly.com/rooms/guided-lab-identifying-a-poisoned-model-artifact
Last updated: 2026-10-13

| | |
|---|---|
| Level | Intermediate |
| Category | Cybersecurity |
| Lessons | 3 |
| Estimated time | 30 min |
| XP | 100 |

This guided lab provides a practical, step-by-step walkthrough of identifying a poisoned model artifact. You will use white-box and black-box detection techniques, including STRIP and Activation Clustering, to uncover a hidden backdoor in a seemingly benign fine-tuned model before it reaches production.

## Lessons

1. Lab Setup and Target Model Overview (30 min)
2. Scanning the Model File for Malicious Payloads (30 min)
3. Confirming and Documenting the Poisoning Vector (30 min)

## Related

- [Learning path: Data and Model Poisoning](https://codelivly.com/learning-paths/data-model-poisoning)
- [All rooms](https://codelivly.com/rooms)