Stage 1 · Lesson 2 of 7
Trust boundaries and untrusted input
A is a line in a system where data or control passes between parts that deserve different levels of trust: the internet and your server, your server and a third-party , a user and an admin. Anything that crosses a boundary from the less trusted side is untrusted input and must be checked before it is used.
Most security bugs are a missed boundary: data treated as safe because it came from "inside". In AI systems there is an extra rule that this roadmap repeats often: the model sits on a boundary too. Its output can be shaped by anyone who can put text in front of it, so it is untrusted input to the next component. Draw boundaries as dashed lines on your diagram.
Common trapdrawing the model inside the trusted zone because your own company runs it.
A question to check yourself, an AI practice session and 1 cheat sheet entry come with the full roadmap. Unlock
Resources for this stage
Start with· follow one of these
- RoadmapFreeSefism Cybersecurity Roadmap
Sefism · Beginner
Sefism's general security roadmap, for any foundation in this stage you want to study in more depth
- CourseFreeProfessor Messer's Security+ (SY0-701) Course
Professor Messer · Beginner
Free videos covering the core security concepts, from access control to cryptography
Practise
Keep handy
Also worth your time
The four questions, diagrams and STRIDE in one practical page
- roadmap.sh: Cyber SecurityGo deeper
A map of the wider security field, so you know where AI testing sits