training / 2026
AI Security Foundations
Training activity focused on security boundaries around AI applications, prompt injection and tool-using systems.
01 / Activity overview
What this activity covered.
A No Breach Training Hub activity focused on understanding how traditional application-security thinking extends into modern AI-enabled systems.
Published highlights
AI application attack surfaces
Prompt-injection concepts
Tool-use security boundaries
Application-security context
02 / Published record
The security perspective behind the activity.
Security beyond the model
AI-enabled applications introduce security questions that extend beyond model behavior alone. The surrounding application, data sources, external tools and authorization model all contribute to the attack surface.
The activity is structured around understanding those boundaries rather than treating AI security as an isolated discipline.
Practical security perspective
Participants examine how prompt manipulation, tool invocation and application permissions can interact.
The objective is to connect emerging AI-security concepts to security principles already familiar from application and API testing.
03 / Training context
Continue into the current AI Security programme.
This page documents a published activity. The Training Hub contains the current programme structure, learning objectives and course information.