A one-prompt attack that breaks LLM safety alignment
As LLMs and diffusion models power more applications, their safety alignment becomes critical.
-
-
Detecting backdoored language models at scale
We’re releasing new research on detecting backdoors in open-weight language models and highlighting a practical scanner designed to detect backdoored models at scale and improve overall trust in AI systems.
