WEB LLM ATTACKS

Summary from the TLDR InfoSec newsletter

This blog post goes through what LLM attacks are and shares examples of how to execute them. Web LLM attacks can retrieve protected data, trigger harmful actions via APIs like SQL injection, and attack systems querying the model. It resembles exploiting SSRF vulnerabilities - abusing a server-side system to attack components not directly accessible.