Information classification
Distinguish public, internal, personal, sensitive, confidential, proprietary, privileged, security-sensitive, and culturally governed information.
Unit overview
Unit 8 examines privacy, security, and data protection across the complete AI workflow. Risks can arise when information is collected, entered into a prompt, retrieved from a source, retained in memory or logs, shared through integrations, reused for another purpose, exposed through an attack, or delivered to an unauthorized person. You will apply purpose limitation, data minimization, source control, least privilege, retention limits, incident response, and accountable stopping through the author-approved PROTECT framework.
Permissions and security
Control who and what may read, retrieve, modify, send, publish, transact, delete, or retain information.
Incidents and stopping
Prepare detection, containment, reporting, correction, notification, fallback, suspension, and stop conditions.
Central idea: Access to information does not establish authority to use it. Responsible AI use requires a legitimate purpose, minimum necessary information, appropriate permissions, verified controls, and a workable response when something goes wrong.
Unit learning outcomes
By the end of Unit 8, you should be able to:
- Distinguish privacy, confidentiality, security, data protection, consent, authority, purpose limitation, minimization, retention, and least privilege.
- Classify information provisionally as public, internal, personal, sensitive, confidential, proprietary, privileged, security-sensitive, or culturally governed.
- Create an information inventory for an AI-supported workflow.
- Identify the minimum information necessary for a legitimate purpose and remove unnecessary identifiers or detail.
- Distinguish technical access from permission, consent, legal authority, professional authority, and community authority.
- Identify information flows among users, interfaces, models, retrieval sources, tools, integrations, providers, logs, and reviewers.
- Assess retention, memory, secondary use, cross-context reuse, disclosure, and vendor-dependency risks.
- Recognize prompt injection, malicious instructions, data exfiltration, insecure integrations, excessive permissions, and unsafe external actions.
- Apply least privilege to read, retrieve, draft, modify, send, publish, transact, control, and delete permissions.
- Design incident detection, containment, reporting, correction, notification, recovery, and evidence preservation.
- Apply the author-approved PROTECT framework to a bounded workflow.
- Extend Assignment 3 with privacy and security analysis while continuing Assignment 1.
Unit 8 lessons
Complete the five lessons in order. The filenames are local to the dedicated Unit 8 folder.
8.1 Information Classification and Purpose Limitation
Classify information, identify legitimate purposes and authority, and distinguish information availability from permission to use.
8.2 Data Minimization, Retention, and Memory
Remove unnecessary identifiers, limit context, evaluate retention and memory, and prevent unauthorized secondary use.
8.3 Permissions, Integrations, and Least Privilege
Map information flows and restrict read, retrieve, modify, send, transact, control, and delete permissions.
8.4 Prompt Injection, Security Threats, and Incidents
Recognize malicious instructions, unsafe tool use, exfiltration, insecure dependencies, and the need for detection and containment.
8.5 PROTECT Workflow and Stop Conditions
Apply PROTECT to purpose, information, authority, retention, permissions, security, incidents, escalation, fallback, and stopping.
Folder convention: Keep
unit-08.html and t-01.html through t-05.html in the same dedicated Unit 8 Moodle folder.Embedded Unit 8 study guide
- Inventory
Identify information, sources, people, systems, and flows. - Minimize
Retain only necessary and authorized information. - Protect
Limit permissions, detect threats, and verify controls. - Respond
Prepare incidents, escalation, fallback, and stopping.
Suggested study sessions
- Retrieval and information scan.
Review FAIRER, read Chapter 10, and inventory information in the Assignment 3 case. - Lesson 1.
Classify information and identify purpose, consent, and authority. - Lesson 2.
Minimize context and evaluate memory, logs, retention, and secondary use. - Lesson 3.
Map flows and create a least-privilege permission matrix. - Lesson 4.
Analyze a supplied prompt-injection or exposure scenario and design incident response. - Lesson 5 and Assignment 3 development.
Apply PROTECT, revise the case analysis, and update Assignment 1.
Evidence to retain
- Information inventory and provisional classifications
- Purpose-and-authority record
- Initial and minimized context
- Retention and memory analysis
- Information-flow map
- Least-privilege permission matrix
- Threat and incident-response analysis
- PROTECT review
- Assignment 3 evidence and Assignment 1 notes
Show the privacy and security review sequence
- Define the legitimate purpose.
- Inventory information, sources, systems, recipients, and flows.
- Classify information and identify applicable authority.
- Minimize collection, context, retention, and access.
- Restrict permissions through least privilege.
- Verify provider, interface, integration, and storage controls.
- Test supplied threat scenarios safely.
- Prepare incident detection, containment, reporting, correction, and recovery.
- Provide human escalation, fallback, and remedy.
- Stop when authority, controls, evidence, or response capacity is inadequate.
Assessment milestones
Assignment 3: Responsible AI Case Analysis
Weight: 20%
Started: Unit 7, Week 5.
Develop now: Add information classification, minimization, authority, permissions, security, incident response, and PROTECT.
Recommended submission: End of Module 3, Week 6.
Official Moodle due date and time: ____________________
Assignment 1: Journal continues
Reflect on information assumptions, privacy, permissions, memory, security, incidents, and how PROTECT changed personal boundaries.
Recommended final submission: End of Module 6, Week 11.
Official Moodle due date and time: ____________________
Use safe case materials: Analyze synthetic, public, supplied, or explicitly authorized information. Do not reproduce actual confidential or sensitive information in the assignment.
Show the Unit 8 Assignment 3 record
- Legitimate purpose and information authority
- Information inventory and classification
- Minimum necessary information
- Sources, recipients, providers, and flows
- Retention, memory, logs, and secondary use
- Permissions and least privilege
- Security threats and dependencies
- Incident response, escalation, fallback, and stopping
- PROTECT findings and residual risk
Required learning activities
| Activity | Purpose | Product | Time |
|---|---|---|---|
| Information inventory | Identify information, source, subject, recipient, system, purpose, and authority. | Inventory table | 35–50 min |
| Classification exercise | Classify supplied information and record uncertainty or required specialist review. | Classification record | 30–45 min |
| Minimization redesign | Remove unnecessary identifiers and detail while preserving the legitimate purpose. | Before-and-after context | 35–50 min |
| Permission matrix | Limit read, retrieve, modify, send, transact, control, and delete authority. | Least-privilege matrix | 35–50 min |
| Threat scenario | Recognize injection, exfiltration, unsafe integration, and excessive permission risks. | Threat analysis | 35–50 min |
| Incident response | Design detection, containment, reporting, correction, notification, and recovery. | Response pathway | 40–55 min |
| PROTECT analysis | Integrate privacy, security, authority, incidents, fallback, and stopping. | Structured review | 45–60 min |
Participation options
Option A: Approved AI
Use an approved system only with synthetic, public, supplied, or explicitly authorized information and no external actions.
Option B: Supplied records
Use supplied prompts, information inventories, flow maps, permission records, threats, and incidents.
Option C: Non-AI analysis
Complete the same information, permission, threat, incident, and PROTECT analysis without an AI platform.
Responsible-use boundaries and support
Boundaries
- Do not upload personal, sensitive, confidential, proprietary, privileged, security-sensitive, or culturally governed records.
- Do not test prompt injection, data exposure, or security weaknesses against live systems.
- Do not connect email, files, calendars, payments, accounts, location, devices, or external-action tools.
- Do not assume deletion, privacy, memory, retention, or training settings without verification.
- Do not provide legal, privacy, security, or professional assurances beyond your authority.
Support and accessibility
- Use supplied records when live AI use is unsuitable or unwanted.
- Provide text alternatives for information-flow and permission diagrams.
- Contact the course coordinator for case and assignment questions.
- Seek qualified privacy, security, legal, professional, Indigenous, or community guidance where applicable.