AI researcher Andrej Karpathy, who co-founded and formerly worked at OpenAI, has joined Anthropic. Karpathy is working on pre-training Claude, one of the most expensive phases in building frontier models.
His appointment comes after a brief stint with Eureka Labs, a startup dedicated to applying AI assistants to education. Karpathy is known for his deep understanding of both LLM theory and large-scale training practice, making him an asset for Anthropic's research team.
Anthropic has also hired Chris Rohlf to its frontier red team, where he will help stress-test advanced AI models against severe threats. With over 20 years of experience in cybersecurity, Rohlf’s expertise could be crucial as AI becomes more integrated into our daily lives.
Karpathy's move is a clear indication that Anthropic believes in the importance of research and development, rather than just relying on compute power to stay competitive with giants like OpenAI and Google. His passion for education may also see him return to this field at some point.
