Securing AI Features in Mobile & Web Apps

About this course

Apps now embed language models in chat, summaries, copilots and autonomous agents, and each one opens a new attack surface. This defensive course maps that surface with the OWASP Top 10 for LLM Applications. You study direct and indirect prompt injection, jailbreaking and guardrail bypass, and insecure output handling where model text reaches HTML, SQL or a shell. You learn how sensitive data leaks, how excessive agency turns an injection into a real action, and how to secure tool-calling and MCP servers, RAG pipelines and vector databases. A lesson on model denial-of-service and leaked API keys covers cost abuse, and the course closes with layered defence in depth. Every lesson has a challenge and quiz, with a practical exam and certificate. Assess only apps you are authorised to test.