Detecting backdoored language models at scalebig tee tech hubFebruary 4, 2026 Today, we are releasing new research on detecting backdoors in open-weight language models. Our research highlights several key properties of…
