Secure Development, lesson 7
AI features and AI-written code
Treat model output as untrusted input, and review AI-written code.
Your team now ships features that call a language model, an AI system that writes text by predicting what comes next, and a model writes a growing share of your code. Neither needs a new process. You'll run both through this module's requirements, review, tests and programme. You need lessons 01, 02 and 06, and Security Foundations' "The risks of AI systems".
Part of Pro
In this lesson
- Checks around a model's output
- Reviewing a change a model wrote
- AI tools in the programme
Cybersecurity comes with Pro: every lesson in every module.
Secure Development overviewSpotted a mistake? Tell us.