Series: AI security

A four-part series on securing your AI assistants, agents and applications: our methodology, the top ten risks, governance in Québec and Canada, and AI penetration testing. To wrap up, a quiz to test your knowledge.

Methodology · 1/4

Our methodology for securing an AI system

How we assess the security of a generative AI assistant, agent or application: inventory, threat modelling, controls, testing and deliverables.

Read the article
Risks · 2/4

The top ten risks of generative AI applications

The OWASP Top 10 for LLM Applications 2026 explained in plain language, with concrete examples and the measures that actually reduce the risk.

Read the article
Governance · 3/4

AI governance: frameworks and obligations in Québec and Canada

NIST AI RMF, ISO/IEC 42001, Law 25, Bill C-36 and the EU AI Act: what applies to your organization and where to start.

Read the article
Penetration testing · 4/4

Penetration testing an AI system, step by step

Prompt injection, jailbreaks, data exfiltration and tool abuse: how an AI penetration test runs, what gets tested and what you receive.

Read the article
Quiz · 5 questions

Test your AI security knowledge

Five questions on AI security: prompt injection, agents, poisoning, governance and Law 25, with explained answers.

Take the quiz

Want your Microsoft 365 tenant audited?

We apply this methodology in your environment, read-only, and hand you a prioritized action plan.

Contact us