AI security testing

How we work

We agree the engagement scope and adapt the activities to your environment, obligations and objectives.

1. 01
   Agree authorised AI applications, agents, data and test boundaries.
2. 02
   Map model, retrieval, tools, trust boundaries and sensitive information flows.
3. 03
   Test scoped scenarios such as prompt injection, data exposure and unsafe tool use.
4. 04
   Validate and prioritise findings with supporting evidence.
5. 05
   Review remediation and retest agreed findings.

What you can expect

A documented scope, relevant findings or gaps, agreed actions and a clear record of ownership. Exact deliverables and timing are agreed for each engagement.

What we need to get started

Your objective, applicable locations or systems, current position, key stakeholders and any target date. Please avoid sending confidential information in the initial enquiry.