Resources
AI Agent Security Engineering Insights
Short explanations of the control decisions behind agent security.
Read the insights
Enforcing guardrails at the action
A model’s intent is not an authorisation decision.
Evidence for an AI Act Article 15 discussion
Establish legal scope before selecting technical evidence.
Human approvals that carry useful context
An approval should describe the action it authorises.
Review tool changes before expanding trust
A previously reviewed tool can change its behaviour.
Review the supplier agent’s actual deployment
A supplier’s general security evidence may not cover the actions you enable.
Read the test design before the headline score
Results depend on what was tested and how.
Start with a clear scope
Tell us which systems, actions and review requirements are in scope. We will discuss the work, responsibilities and deliverables before you commit.