Artificial Intelligence (AI) · Advanced level

Advanced level · Module 4 of 6 · 22 guided hours

AI Safety, Security and Testing

Protect AI systems from prompt injection, data leaks and misuse, test them adversarially and monitor them in production.

Learning objectives

  • Defend against prompt injection and data exfiltration
  • Red-team AI systems and fix findings
  • Monitor AI systems in production

Topics

  • prompt injection (direct and indirect)
  • data exfiltration through tools
  • content safety and misuse
  • red-teaming
  • security controls
  • production monitoring and incident response

Lessons

  1. Lesson 1 Prompt Injection and Data Leaks

    Understand how attackers hijack AI systems through instructions in inputs and documents, and defend against it.

    prompt injectionindirect injection via documentsdata exfiltrationdefence in depth

  2. Lesson 2 Misuse and Content Safety

    Prevent AI systems from being used to cause harm or produce harmful content.

    misuse scenarioscontent filteringusage policiesabuse monitoring

  3. Lesson 3 Red-Teaming AI Systems

    Systematically attack your own system to find failures before users or attackers do.

    red-team planattack categoriesfindings and severityretesting

  4. Lesson 4 Monitoring and Incident Response

    Monitor AI systems in production and respond when things go wrong.

    quality signalsuser feedbackdrift and changes in providers' modelsincident response

Practical activity

Red-team the capstone system — including injected instructions hidden in documents — fix the findings and set up monitoring and an incident procedure.

Expected outcomes

  • More secure AI systems
  • Testing and monitoring habits

Assessment

Red-team report with fixes and evidence, plus the monitoring plan.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.