Apollo Research to Scale Fine-Tuning of AI Monitors

Marius Hobbhahn serves as CEO and co-founder of Apollo Research, a group dedicated to AI safety through model evaluations and monitoring. In a recent post he noted that monitoring involves far too many critical tasks. The main constraint remains a lack of additional personnel. He identified staffing shortages as the chief blocker preventing faster progress on these monitoring initiatives. Apollo Research intends to launch fine-tuning of monitors at larger scale soon. Early tests of the method have delivered strong outcomes. Hobbhahn expressed the view that this development may hold substantial value for managing schemers in advanced AI models.