Phase 18: Ethics, Safety & Alignment
AI From Scratch/Lesson 08/~60 minutes

In-Context Scheming in Frontier Models

Meinke, Schoen, Scheurer, Balesni, Shah, Hobbhahn (Apollo Research, arXiv:2412.04984, December 2024). Tested o1, Claude 3.5 Sonnet, Claude 3 Opus, Gemini 1.5 Pro, Llama 3.1 405B on agentic scenarios where the in-context prompt creates a co...

LearnNo prerequisites
Loading lesson page...