blog posts U of I researchers show how large language models can auto-hijack websites Feb 19, 2024 2:30 pm by The Register (London)Illinois in the News Share on Facebook Tweet Email Images The Register (Feb. 17) 'What really worries me about future highly capable models is the ability to do autonomous hacks and self-reflection to try multiple different strategies at scale,' says Illinois Professor Daniel Kang. Share on Facebook Tweet Email
