For training dataFor inferenceAI Hacking MethodsAdversarial AttackAI JailbreakPrompt Injection AttackLLM extraction attackAI Reward Hacking Universal and Transferable Attacks on Aligned Language Modelshttps://llm-attacks.org/?fbclid=IwAR2fNkjoOdg8qIgNXEPIvyLjboYr4My4NN9Bx89J-Yx7UElSTyKT89_3JeEPoisonGPT: How we hid a lobotomized LLM on Hugging Face to spread fake newsWe will show in this article how one can surgically modify an open-source model, GPT-J-6B, and upload it to Hugging Face to make it spread misinformation while being undetected by standard benchmarks.https://blog.mithrilsecurity.io/poisongpt-how-we-hid-a-lobotomized-llm-on-hugging-face-to-spread-fake-news/
Seonglae ChoSeonglae ChoSeonglae ChoSeonglae ChoSeonglae ChoSeonglae Cho
