Influence and cyber operations an update october 2024

Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training

Technical Report: Large Language Models can Strategically Deceive their Users when Put Under Pressure

Role-Play with Large Language Models