confused deputy ai agent
-
MITRE ATTACK V20 AI Techniques: How to Test Them
MITRE ATTACK v20 AI techniques cover prompt injection, AI-enabled exfiltration and result poisoning. See how to turn each into a test that proves impact.
-
What Is an AI Agent? A Security-First Breakdown
What is an AI agent? See how the model, tools and permissions loop really works, where agents break, and how to check one before it goes live.
-
OpenAI Misaligned Models: From Reports to Tests
OpenAI misaligned models made headlines, but what should you actually test? See how to turn the disclosed cases into tests you can run and prove.
-
LLM Penetration Testing Checklist: Proof Over Output
An LLM penetration testing checklist built around evidence: what to map, what to attack, and how to prove impact. See the checks in full.
-
AI Coding Agent Security: 7 Attack Paths to Test
AI coding agent security tested properly: seven attack paths from prompt injection to shell abuse, how to prove each one, and the controls that hold. Read on.
-
AI Red Teaming Methodology: From Scope to Proof
A working AI red teaming methodology: map permissions first, test five attack classes in order, and prove consequence. See how to structure it.
-
MCP Server Prompt Injection: Proving the Blast Radius
MCP server prompt injection is easy to demo and hard to prove. See how to test to consequence, scope tokens and write findings that hold up.