GenAI

A large database full with the latest cyber intelligence info like threat reports, news articles and more that can easily be queried using an LLM? Sounds expensive but can be achieved with open-source components for free!

This post is about safety alignment bypasses like prompt injections or jailbreaks, that make LLms ignore their guardrails.

This new series covers security risks in AI systems in depth. From prompt injections, to supply chain attacks and risks of agentic AI.