Speaker: Marco Di Filippo

 

What happens when cybercriminals deliberately use Large Language Models (LLMs) for attacks? Modern AI systems enable the generation of advanced malware, the analysis of code fragments, and the simulation of security mechanisms in order to identify and exploit vulnerabilities, including potential zero-day vulnerabilities.

 

The presentation also demonstrates how techniques such as prompt injection, agentic chain manipulation, and manipulated AI interactions can be used to initiate attacks against AI infrastructures themselves.

 

This talk clearly ...