Week 10 · Phase 4: Rules & defenses

Enterprise AI controls: SSO, DLP, CASB/SSE

The real job: rolling out ChatGPT, Claude and Gemini safely at work.

Time this week: 8–10 hours

Goal for this week
By Sunday you can design controls for an enterprise AI rollout, write DLP rules that catch sensitive data in prompts, and triage shadow AI.
Why it matters for AI security
This is the most common AI security task in companies today: staff want AI tools, and security must allow them safely. That means the right plan, SSO, data rules, DLP, and watching for 'shadow AI'.

Start here: this week's labs

DLP Policy Lab: stop leaks into AI prompts
Write DLP rules (keywords and patterns), choose block, warn or redact, and test them against 16 real-style prompts. Score your catches and false alarms.
Open lab
Shadow AI Lab: CASB triage
A CASB dashboard shows the AI apps staff are using. Decide allow, limit or block for each, and write the reason.
Open lab

Where enterprise AI controls sit

Employeewrites a prompt
SSO + MFAwho are you?
DLPsensitive data? block / redact
CASB / SSEapproved AI app?
Enterprise AIChatGPT / Claude / Gemini
Audit logsrecord for investigations

Key ideas

Enterprise plans

Business versions like ChatGPT Enterprise, Claude for Work (Team/Enterprise) and Gemini for Google Workspace offer SSO, admin controls, audit logs, and contract terms about not training on your data. Always check the vendor's current terms.

Shadow AI

Staff using AI tools IT doesn't know about, often free personal accounts. Data pasted there may leave company control.

DLP

Data Loss Prevention. Rules that spot sensitive data (card numbers, IDs, secrets, customer data) and block, warn or redact it before it leaves.

CASB / SSE

Cloud Access Security Broker / Security Service Edge. Tools that see which cloud apps (including AI apps) people use and apply policy: allow, block, or limit.

Audit logs

Records of who used the AI tool, when, and (depending on plan) what was sent. Needed for investigations.

Data classification

Labels like Public, Internal, Confidential, Restricted. DLP and AI rules are based on these labels.

Day-by-day plan

Tick each day when you finish it. Your ticks are saved in this browser.

Mini game: Which control fixes it?

Practice outside this course (free)

ChatGPT Enterprise: security & privacy
OpenAI's enterprise privacy page.
Anthropic Trust Center
Security and compliance info for Claude.
Google Workspace: Gemini privacy hub
How Gemini handles Workspace data. Search 'Gemini Workspace privacy hub' if the link moves.
Microsoft Purview DLP intro
Free, clear explanation of DLP.

These links need the internet. Everything else on this page works offline.

Exercises & worksheet

Exercise deck (PPTX) Worksheet (DOCX)

Weekly quiz

Finished the week?
Mark it done to fill your progress bar on the course home.
← Week 9: AI governance: NIST, ISO 42001, EU AI Act Week 11: AI red teaming & tools →