AI › Module 10 › Lesson 2
Lab Path
AI Security literacy path on `$AI_LAB` — lab setup to mentor handoff with ethics gates at every station.
Visual · t32_lab_path
Lab path = visual AI security chain on YOUR $AI_LAB. Original Cyberlium.
Opening
Draw the path from lab scope to mentor handoff — with stop signs at every jailbreak kit shortcut and stranger app test lane.
Document literacy path: `$AI_LAB` setup → LLM basics → injection → poisoning/threats → LLM app security → secure dev → governance/red team → findings/disclosure → capstone. Ethics gates: jailbreak weaponization, stranger app testing, fraud deepfake kits, prod LLM without RoE, public bypass publication — refused paths labeled red. UTC milestones and artifact IDs per station. Mermaid or sketch chmod 600. Honest N/A if brief skipped live eval — labeled. Defender breakpoint: where RoE or YOUR apps discipline prevented out-of-scope test.
1. Path stations
Setup, LLM basics, injection, poisoning, threats, LLM apps, secure dev, governance, findings, handoff — ten minimum stations.
Each station links evidence index row and module lesson ID.
Command guide
Try these commands — Path stations
═══ 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/ NIST AI RMF — https://www.nist.gov/itl/ai-risk-management-framework OpenAI safety — https://openai.com/safety
═══ INSTALL ═══
Linux (Debian/Ubuntu):
Command — copy this
sudo apt install curl
macOS: Built-in
Windows: Built-in (PowerShell: Invoke-WebRequest)
═══ LINUX / macOS ═══
Command — copy this
export AI_LAB=${AI_LAB:-$HOME/cyberlium-lab/t32-ai}
echo '=== Capstone path: mindset → LLM basics → injection → poisoning → threats → OWASP apps → secure dev → governance → findings ==='
grep -E 'LLM0|injection|echo_bot' "$AI_LAB/owasp/llm-top10-index.txt" "$AI_LAB/injection/direct-injection-named.txt" | head -5
grep -E 'Detection|HARD BAN' "$AI_LAB/threats/ai-phishing-detection-signals.md" "$AI_LAB/threats/deepfake-detection-signals.md" | head -4
curl -sS https://owasp.org/www-project-top-10-for-large-language-model-applications/ | head -5Primary tools to practice this lesson: curl, grep. Reference sites: OWASP LLM Top 10 (https://owasp.org/www-project-top-10-for-large-language-model-applications/); NIST AI RMF (https://www.nist.gov/itl/ai-risk-management-framework); OpenAI safety (https://openai.com/safety). Run every command in the box — install first, then the usage lines — only on YOUR lab / program scope.
2. Ethics overlay
Red stops: jailbreak kits, stranger apps, fraud deepfakes, prod without RoE, public bypass.
RoE branch always green start — `$AI_LAB` on separate labeled branch.
3. Integrity breakpoint
Mark where scope check or YOUR apps rule prevented out-of-RoE test.
One paragraph uplift from YOUR `$AI_LAB` path.
4. What you ship: lab path diagram
AI security chain with ethics gates + UTC milestones — chmod 600.
5. What you record before the next lesson
Lab path diagram 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
Lab path includes stranger LLM API injection shortcut lane.
Right
Lab path documented for $AI_LAB. Next: Evidence Pack.
Mission: lab path diagram
1) Map ten stations. 2) Add six ethics gates. 3) UTC milestones per station. 4) Link evidence index IDs.
Stuck? Ask Cyberlium AI Mentor
Ask Mentor: “Diagram vs numbered list for capstone?”
Knowledge Check
APPLY: Lab path documents:
Multiple choice
Knowledge Check
APPLY: True or False: Stranger app shortcut OK on path.
True or False
Knowledge Check
APPLY: Path ends at:
Multiple choice