Tag
4 items filed under #Handpicks.
Research documenting how adversarially crafted tool outputs can establish false premises in language models, leading to compounding misalignment and harmful outputs including defamatory content.
Exploring how social engineering techniques can be used to manipulate LLMs into bypassing their safety measures
A simulated test reveals how a flawed rule and authoritative pressure can lead an AI to make a decision with severe ethical consequences, highlighting crucial areas for AI safety research.
Exploring the Latest AI Models I've Jailbroken