COURSE 19Intermediate8 IN-DEPTH LESSONS

Safety, Security, Bias and Alignment

Architecting resilient systems: prompt injection defense, sandboxing, data minimization, algorithmic fairness, and secure software lifecycles.

Start Lesson 1: Bias, fairness, and accountability →← All 29 Courses

Course Syllabus & Units

Unit 1

Bias

1 lessons
1.1L1

Bias, fairness, and accountability

AI systems can treat people **unequally** because of skewed data, flawed labels, or design choices—even when nobody typed a hateful rule. **Fairness** work tries to measure and reduce unjust harms. **Accountability** asks who is responsi...

Read lesson →
Unit 2

Alignment

1 lessons
2.1L1

Alignment and safety

**Alignment** (in everyday AI talk) means making systems behave in line with intended human goals and constraints—not just maximizing a narrow score. **Safety** covers preventing harms from mistakes and misuse.

Read lesson →
Unit 3

Security

1 lessons
3.1L1

Security basics for AI apps

AI apps inherit normal app security **plus** new failure modes: prompt injection, tool abuse, and data leaking into prompts/logs.

Read lesson →
Unit 4

Privacy

1 lessons
4.1L1

Privacy and data minimization

Collecting and sending only what you need—**data minimization**—and protecting what remains: access control, retention limits, and careful vendor choices.

Read lesson →
Unit 5

Moderation

1 lessons
5.1L1

Content moderation systems

Systems that detect and act on policy-violating content—human+ML hybrids with appeals. Literacy on design tradeoffs, not a how-to evade.

Read lesson →
Unit 6

Sdl

1 lessons
6.1L1

Secure SDL for AI

Folding AI-specific risks into the **secure software development lifecycle**: threat model → design → implement → test → deploy → respond.

Read lesson →
Unit 7

Incident

1 lessons
7.1L1

Incident response for AI

How teams respond when AI features leak data, cause harm, or get abused—playbooks, severity, comms, and fixes.

Read lesson →
Unit 8

Supply

1 lessons
8.1L1

Model supply chain basics

Treating models, adapters, datasets, and prompts like software dependencies—with integrity checks and vendor trust.

Read lesson →