Prompt Injection & Jailbreak Defence
Learn how people trick AI systems into ignoring their instructions by doing it in a safe practice app, then add defences one at a time and measure how many attacks still get through.
Modules
6
Verified
0
In Progress
0
What you walk away with
- ✓A shareable portfolio URL with your project walkthrough
- ✓Module-by-module timeline of everything you built
- ✓All the code — pipelines, guardrails, deployment configs
- ✓Production patterns documented on your profile
An intermediate course for developers, AI builders and security engineers. You run 40 described attack records against Ask Corri, a made-up insurer's help assistant: direct overrides, role-play requests, requests for its instructions, tool misuse, data leakage, and hidden instructions planted in documents, uploads and claim notes. Then you switch on five defences built into the app (secrets out of the instructions, documents and tool results marked as data, tool permissions checked in code, a person approving refunds, and an output check), measure each alone and in layers, write a rule for each tool, and finish with a repeatable attack test that runs before every change. Defensive and educational: no attack wording, and attacks only run against the course's own practice app. No defence is complete.