November 2026
-

NCSC AI Ransomware Warning: 7 Tests to Run
The NCSC AI ransomware warning, explained plainly, plus seven tests that prove what a hijacked assistant or agent could reach. See where to start.
-

Confused Deputy AI Agent: 6 Tests That Prove It
Confused deputy AI agent risk explained: six tests that prove when an agent acts on its own authority, plus the evidence that holds up. See how to run…
-

How to Pentest an AI Agent: A 7-Step Runbook
How to pentest an AI agent end to end: map permissions, find hidden input channels, abuse the tool layer and prove findings twice. See the full runbook.
-

AI Agent Prompt Injection: 6 Blast Radius Tests
AI agent prompt injection is judged by what the agent did, not what it said. Six tests that map tool abuse, exfiltration and reach. See how to run…
-

NCSC Annual Review 2026: The AI Risks to Test
NCSC Annual Review 2026 flags AI as a growing risk. See how to turn its findings into concrete tests for your own LLM apps and agents.
-

AGENTS.md Prompt Injection: How to Test and Contain
AGENTS.md prompt injection lets a cloned repo steer your coding agent. See how to test the file, prove the blast radius and scope permissions.


