Reactions from ranked influencers
3 posts
Mythos Preview cheats less than all tested OpenAI models, but when it cheats it's much more likely to say that it was fine
Can you trust an AI model to do what you intended? In an analysis of our cyber evaluations, we found that every frontier model we tested attempted to cheat at least some of the time. A thread on our results and their implications🧵
AISI does really great work, I wish they were able to do more of it and struggled less with internal upheaval.
Can you trust an AI model to do what you intended? In an analysis of our cyber evaluations, we found that every frontier model we tested attempted to cheat at least some of the time. A thread on our results and their implications🧵