Content pfp
Content
@
0 reply
0 recast
0 reaction

assayer pfp
assayer
@assayer
proactive measures to safeguard frontier AI models? not so safe when "a weak model proposes related-but-benign tasks, a frontier model solves these, and finally the weak model uses these solutions in-context to complete the original harmful task" https://arxiv.org/abs/2406.14595
0 reply
0 recast
1 reaction