AI Resource Library

Every AI resource we have written, in one place.

Strategy, implementation, security, governance, and the latest ASD guidance on agentic AI. Browse by what you are trying to do, not by what we are trying to sell.

Last updated:
Latest guidance: May 2026

ASD and Five Eyes: Careful adoption of agentic AI services

In May 2026 the Australian Signals Directorate and its Five Eyes partners published joint guidance on the careful adoption of agentic AI services. The guidance followed testing that showed advanced AI models can autonomously reason about objectives, adapt to changing circumstances, and combine multiple technical actions into sophisticated attack sequences.

Least privilege

Limit agent permissions to the minimum required to perform approved tasks.

Human oversight

Maintain human oversight and approval for high-impact or sensitive actions.

Continuous monitoring

Monitor agent behaviour, decisions, and tool usage with alerts for anomalous activity.

Comprehensive logging

Audit agent actions so they can be reviewed, traced, and reversed if needed.

Red teaming

Regular adversarial testing before and during deployment.

Progressive deployment

Start with low-risk use cases. Increase autonomy only as assurance matures.

Not sure where to start?

Book an AI strategy session.

Fifteen minutes with one of our engineers. We will tell you honestly whether AI is worth pursuing for your business right now, and if so, where to start.