LLM Goat
This project is an open source, deliberately vulnerable environment created by SECFORCE to learn about LLM-specific risks based on the OWASP Top 10 for LLM Applications.
Are you Largely Learning Motivated?
Bang your head against real-world LLM vulnerabilities.
We designed a series of 10 challenges simulating real-world vulnerabilities so you can safely test and understand risks associated with large language models.
Solutions to the challenges will be
published here soon, stay tuned!
07
System Prompt Leakage
08
Vector and Embedding Weaknesses
09
Misinformation
10
Unbounded Consumption
No goats were harmed during the creation of this tool…
but we can't say the same for our own LLMs, unfortunately. 😅
