Kyle O’Brien
Co-Founding Engineer
Kyle is a technical AI safety researcher and former ERA fellow, previously contributing as a scientist and engineer at EleutherAI and Microsoft. His research focuses on preventing and/or removing dangerous capabilities from LLMs, with a specific focus on pretraining interventions and open-weight model safety. Kyle's research has been accepted to ICLR, cited by OpenAI and Anthropic, and covered by The Washington Post and Fortune.