notesum.ai

Published at December 6

Frontier Models are Capable of In-context Scheming

cs.AI
cs.LG

Released Date: December 6, 2024

Authors: Alexander Meinke1, Bronson Schoen, Jérémy Scheurer, Mikita Balesni, Rusheb Shah, Marius Hobbhahn

Aff.: 1Apollo Research

Arxiv: http://arxiv.org/pdf/2412.04984v1