Göktuğ Serdar Yıldırım
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
BrainwashedLLM: Jailbreaking via Automated Adversarial Chain of Thoughts Discovery
According to the abstract, Masoud Poorghaffar Aghdam, Göktuğ Serdar Yıldırım and Sinem Sav describe BrainwashedLLM, a method that trains a small model to write adversarial reasoning traces. Injected as prefixes, the traces are said to steer larger black-box reasoning models to produce harmful outputs.