Misleading instructions
We check whether manipulated requests can change the intended behaviour.
AI / KI
Security for your AI assistant.
We check how a chatbot handles instructions, information and actions, within an agreed scope.
What is included
Confidential information
We check whether the system can expose data or instructions it should protect.
Actions and permissions
Where tools are connected, we check the boundaries of authorised operations.
Scope, authorisation and test limits are agreed before we begin. You receive a report with evidence, priorities and recommended fixes.
Let’s discuss your project.
An initial conversation with no obligation to understand your goals, priorities and next steps.
Let’s talk