Category: AI Security & Governance
Securing AI and LLM systems, plus the frameworks that govern them.
-
Prompt injection is not a bug you patch
In an LLM application the instruction and the data arrive through the same channel. That is an architectural problem, not a filtering one.
-
MITRE ATLAS gives AI attacks a shared vocabulary
ATT&CK made network intrusion discussable. ATLAS is doing the same for machine learning systems.
-
AI governance is an inventory problem before it is a policy problem
You cannot govern AI systems you cannot list. Start with the register, then apply NIST AI RMF and the EU AI Act.