AI › Module 9 › Lesson 4
Lab — Findings
Findings lab on `$AI_LAB` — writeup, severity backlog, responsible disclosure checklist bundled.
Visual · t32_findings_lab
Lab: findings pack from YOUR $AI_LAB eval artifacts. Original Cyberlium.
Opening
Ship professional AI findings artifacts from YOUR authorized eval — writeup, triage, ethics — zero jailbreak kits or stranger app evidence.
From `$AI_LAB` labs Modules 6–8: (1) report cover — scope, RoE, analyst, UTC; (2) one complete finding writeup Module 9-1; (3) prioritized backlog five findings Module 9-2; (4) responsible disclosure checklist Module 9-3; (5) remediation summary stub linking app/eval/governance rows; (6) limitations paragraph (lab vs prod, brief gaps); (7) integrity — `$AI_LAB` only, no stranger app test, no jailbreak kit repro, no prod paste; (8) chmod 600 index.md. Professional tone — mentor-readable. Cross-link LLM app, secure dev, and governance packs.
1. Lab deliverables
Writeup + priority backlog + responsible disclosure — single findings thread.
Limitations paragraph mandatory — honest lab scope.
Command guide
Try these commands — Lab deliverables
═══ TOOLS & WEBSITES ═══ Browse / read these (authorized learning only — stay in YOUR lab / program scope)
OWASP LLM Top 10 — https://owasp.org/www-project-top-10-for-large-language-model-applications/ MITRE ATLAS — https://atlas.mitre.org/ NIST AI RMF — https://www.nist.gov/itl/ai-risk-management-framework
═══ INSTALL ═══
Linux (Debian/Ubuntu):
Command — copy this
sudo apt install python3
macOS:
Command — copy this
brew install python3
Windows: Download https://python.org/downloads/
═══ LINUX / macOS ═══
Command — copy this
export AI_LAB=${AI_LAB:-$HOME/cyberlium-lab/t32-ai}
cat > "$AI_LAB/findings/findings-lab-index.md" <<'EOF'
# Findings Lab Index — YOUR authorized tests
- finding-writeup-template.md (copy per finding)
- severity-triage.md
- responsible-ai-disclosure.txt
- Link injection lab: bot/echo_bot.py test results
- Map each finding to OWASP LLM01–LLM10
## Integrity
- All findings from YOUR lab toys or YOUR staging apps only
EOFCommand — copy this
grep -E 'OWASP LLM|Integrity|YOUR' "$AI_LAB/findings/findings-lab-index.md" ls "$AI_LAB/findings/"
═══ WINDOWS ═══
Command — copy this
Get-Content $HOME/cyberlium-lab/t32-ai/findings/responsible-ai-disclosure.txt | Select-String NEVER
Primary tools to practice this lesson: grep, python3. Reference sites: OWASP LLM Top 10 (https://owasp.org/www-project-top-10-for-large-language-model-applications/); MITRE ATLAS (https://atlas.mitre.org/); NIST AI RMF (https://www.nist.gov/itl/ai-risk-management-framework). Run every command in the box — install first, then the usage lines — only on YOUR lab / program scope.
2. Quality bar
Each finding traceable to Module 6–8 evidence row.
Zero jailbreak kits or stranger app refs in any section.
3. Integrity scan
Scan for stranger refs, bypass kits, prod paste — zero hits.
Practice label if brief requires.
4. What you ship: findings lab pack
Writeup + triage + ethics + index + integrity — chmod 600.
5. What you record before the next lesson
Findings lab pack path.
6. Wrong vs right: stranger SaaS vs YOUR toy LLM
Worked failure — same MSF word, opposite target. Right never needs a café Wi-Fi or classmate laptop.
Wrong
Findings lab includes public jailbreak kit appendix and stranger app scan export.
Right
Findings lab pack on `$AI_LAB`. Next: quiz.
Mission: findings lab
1) Finding writeup complete. 2) Priority backlog five rows. 3) Responsible disclosure checklist. 4) Integrity scan; chmod 600.
Stuck? Ask Cyberlium AI Mentor
Ask Mentor: “Minimum writeups if brief sparse?”
Knowledge Check
APPLY: Findings lab includes:
Multiple choice
Knowledge Check
APPLY: True or False: Hide limitations in findings lab.
True or False
Knowledge Check
APPLY: Finding writeup repro:
Multiple choice