Andrej Karpathy, former OpenAI, joins Anthropic to boost AI

Andrej Karpathy, an iconic figure in artificial intelligence, has recently joined Anthropic, a company specializing in the development of advanced AI models. Karpathy, who co-founded OpenAI and led AI at Tesla, announced his arrival at Anthropic via a post on X, expressing his enthusiasm for the upcoming advancements in the field of large language models (LLMs).

Karpathy began his role at Anthropic this week, where he is focusing on pre-training under the guidance of Nick Joseph. Pre-training is a crucial phase that involves large-scale training sessions, essential for equipping Claude, Anthropic's model, with its foundational capabilities. This stage is recognized as one of the most costly and computationally demanding in the development of cutting-edge AI models.

A spokesperson for Anthropic told TechCrunch that Karpathy will be forming a team focused on leveraging Claude to accelerate pre-training research. Karpathy is one of the few researchers capable of bridging the theory of LLMs and the practice of large-scale training. His recruitment is a clear indication that Anthropic is betting on AI-assisted research to remain competitive against OpenAI and Google.

Before joining Tesla, Karpathy focused on deep learning and computer vision at OpenAI. He left OpenAI in 2017 to lead Tesla's Full Self-Driving (FSD) and Autopilot programs until 2022. After a brief return to OpenAI for a year, he left again in 2024 to found Eureka Labs, a startup dedicated to applying AI assistants in education. Although little information has been shared about the future of this company, Karpathy has expressed his ongoing interest in education, particularly through his online course "Neural Networks: Zero to Hero."

In parallel, Anthropic has also strengthened its team with the arrival of Chris Rohlf, a cybersecurity expert. Rohlf, who has over 20 years of experience in the field, previously worked for Yahoo and Meta. At Anthropic, he will focus on stress testing AI models, a crucial role for assessing their robustness against potential threats. Rohlf has also been a researcher at the Center for Security and Emerging Technologies at Georgetown, where he worked on the CyberAI project.

Rohlf has expressed his belief that AI presents a significant opportunity to enhance cybersecurity, emphasizing the importance of his new role at Anthropic in a context where the security of AI models is increasingly scrutinized.