From OpenAI to Anthropic, Andrej Karpathy Joins Anthropic to Lead Pre-Training Research
Andrej Karpathy joins Anthropic, and the AI world is paying close attention. The renowned researcher and OpenAI co-founder confirmed the move on Tuesday in a post on X, saying he is “very excited to join the team and get back to R&D.” He started this week.
Karpathy is joining Anthropic’s pre-training team under team lead Nick Joseph. Pre-training is the large-scale training process that gives Claude its core knowledge and capabilities, and it is one of the most compute-intensive phases of building a frontier model. According to Anthropic, Karpathy will start a team focused specifically on using Claude to accelerate pre-training research itself, a move that signals the company’s bet on AI-assisted research as a competitive edge over rivals like OpenAI and Google.
Karpathy is widely regarded as one of the few researchers who can bridge the gap between LLM theory and large-scale training practice, making this hire particularly significant. His resume reads like a tour of the most consequential AI projects of the past decade. He focused on deep learning and computer vision at OpenAI before departing in 2017 to join Tesla, where he led the Full Self-Driving and Autopilot programs until 2022. He then returned to OpenAI for a year before leaving in 2024 to found Eureka Labs, a startup dedicated to applying AI assistants to education.
What happens to Eureka Labs now is unclear. Karpathy has shared few updates on it since its launch. He also runs a popular YouTube channel with AI lectures and created the widely followed Neural Networks: Zero to Hero course. He said in his announcement that he “remains deeply passionate about education” and plans to return to that work eventually.
In a separate but related development, Anthropic also brought on cybersecurity veteran Chris Rohlf to its frontier red team, the unit that stress-tests advanced AI models against severe threats. Rohlf brings over 20 years of experience, with prior stints at Yahoo’s elite security group, Meta, and Georgetown’s Center for Security and Emerging Technology.
Both hires point to an Anthropic that is aggressively strengthening both the science and the safety of its models as the race at the AI frontier heats up.
