An app that works is not the same as an app that is safe. AI tools aim to produce code that runs, and they can easily leave behind problems such as exposed API keys, missing login checks, or databases that anyone can read. Before real people use your app, you need to check for these.
This module covers the most common security risks in AI-generated code, how attackers find them, and how to protect your secrets and your users' data. It also shows how to test your app, including using AI to help write tests, so you catch problems early instead of after launch.
Watch the lessons in order, then continue on to Limits & Shipping Real Products.
The main risks of AI-assisted development, from leaked API keys to made-up packages, and how to secure what the AI builds.
Why vibe-coded apps create security problems, especially around access to sensitive data, and what to do about it.
An AI security engineer attacks a live, AI-built app and finds 13 exploitable vulnerabilities, 8 of them high severity.
Why testing matters even more when AI writes the code, and how different kinds of tests catch problems.
What makes AI-generated tests actually useful, for anyone who thinks models only write poor tests.