Today it feels like everyone and their grandmother is finding high severity security bugs with language models. So is this the end of vulnerability research? This talk traces how LLMs got here, what they can do today, and where they still struggle.
Cim has been breaking software since his early teens, on both the offensive and defensive sides. He has competed at Pwn2Own and holds CVEs in browsers and kernels.
For the past eight years his day job has been building machine learning systems across computer vision, natural language, and time series. When not in front of a terminal, he’s usually jumping out of a plane.
