Agents let AI safety share experiments hourly, not just papers monthly

A new approach called xAgents enables AI safety researchers to share experimental results in near real-time, accelerating collaboration beyond traditional monthly paper publications. This method aims to enhance the pace and transparency of AI safety research.