OpenAI co-founder Karpathy joins Anthropic for pre-training R&D

Andrej Karpathy, the renowned AI researcher who co-founded OpenAI and led Tesla's Full Self-Driving and Autopilot programs, has joined Anthropic. Karpathy announced the move on X, stating he is excited to return to R&D at the frontier of LLMs. He will work on pre-training under team lead Nick Joseph, focusing on using Claude to accelerate pre-training research. Pre-training is one of the most expensive and compute-intensive phases of building frontier models. Karpathy's deep expertise in both theory and large-scale training makes him a key hire for Anthropic, signaling the company's bet on AI-assisted research over pure compute to compete with OpenAI and Google.

Karpathy's career includes stints at OpenAI (focus on deep learning and computer vision), Tesla (leading FSD and Autopilot), a return to OpenAI for one year, and then founding Eureka Labs, an AI education startup. He also maintains a popular YouTube channel and teaches an online course, Neural Networks: Zero to Hero. Karpathy says he remains passionate about education and plans to resume that work later. Separately, Anthropic hired Chris Rohlf, a veteran cybersecurity expert from Meta and Yahoo's Paranoids team, to join its frontier red team, which stress-tests advanced AI models against severe threats.