LLM Security
Securing the new wave of Generative AI applications.
Operational Phase
01
Prompt Injection
Tricking the LLM into ignoring instructions (DAN mode).
02
Jailbreaking
Bypassing safety filters to generate harmful content.
03
Data Leakage
Ensuring PII doesn't end up in training data.
04
Indirect Injection
Attacking LLMs via web pages they read.