LevelUp AI × EveryAge AI · EveryAge AI

Day 27: Prompt Injection, Jailbreaks & Boundary Testing (Safe Prompting & Privacy)

EveryAge AI · Phase 2 (Days 16–30): Safe Prompting, Privacy & Creative Co-Design Track

🔊 Voice-Friendly Read-Aloud / Narration Script:

Day 27 (Phase 2 (Days 16–30): Safe Prompting, Privacy & Creative Co-Design): Prompt Injection, Jailbreaks & Boundary Testing. Goal: Understand how adversarial prompts attempt to bypass safety rules and how to keep instructional prompts focused.

Learning Goal: Understand how adversarial prompts attempt to bypass safety rules and how to keep instructional prompts focused.

Key Concepts

Core Principle (Day 27): Understand how adversarial prompts attempt to bypass safety rules and how to keep instructional prompts focused.
Zero-PII Safeguard: Strip all student, child, family, client, and staff identifiers before using any digital tool.
Human-in-the-Loop Verification: Cross-check every factual claim, citation, calculation, and tone against primary sources before adoption.

Hands-On Activity

Inspect 4 fictional examples of off-topic or adversarial prompt inputs and write clear system boundaries.

Practice Prompt Template (Zero-PII)

Explain in age-appropriate terms why a digital tool must refuse requests that ask for private data, unsafe actions, or academic dishonesty.

Human-in-the-Loop Checkpoint: Document the refusal and boundary rules learners must follow.

6 Co-Branded Track Emblem System (K–2 to Adult Workforce)

Privacy & Governance Safeguard: Never enter student, child, staff, family, or confidential workplace PII into unapproved AI tools. EveryAge AI is an independent educational & professional literacy program by EveryAge AI (Inquiries: hello@everyageai.com / oes@everyageai.com).