December 6, 2024news An evaluation of six frontier AI models for in Apollo Research evaluated frontier models for in-context scheming capabilities. We found that multiple frontier models are capable of in-context scheming when strongly […]
