Summary (tl;dr)

Andrej Karpathy, a prominent AI researcher and co-founder of OpenAI, has announced he is joining Anthropic, a competing artificial intelligence lab, where he will lead a new team focused on advancing AI pre-training research using Anthropic's Claude model.

Essential Background

Andrej Karpathy has a distinguished career in the field of artificial intelligence, including being one of the original eleven co-founders of OpenAI and serving as the Director of AI at Tesla, where he led the computer vision team for their Autopilot system. After leaving OpenAI in 2024, he dedicated his time to AI education, establishing Eureka Labs, an AI-native education startup, and creating popular educational content on large language models and neural networks. Karpathy is also credited with coining the term "vibe coding," referring to using AI tools to develop programs with minimal human intervention.

The Full Story

On Tuesday, May 19, 2026, Andrej Karpathy publicly announced via the social network X that he has joined Anthropic. In his announcement, Karpathy expressed his enthusiasm to return to research and development, stating his belief that the next few years in large language models (LLMs) will be particularly formative. At Anthropic, he will join the pre-training team, which is responsible for the extensive training runs that establish the core knowledge and capabilities of their AI model, Claude. Specifically, Karpathy will be building a new team dedicated to using Claude itself to accelerate pre-training research, an initiative that could contribute to the development of "recursive self-improvement" in AI, where AI systems are capable of training or upgrading their successors. This significant hire for Anthropic comes amidst intense competition for top AI talent among leading technology companies.

Why It Matters

Karpathy's move to Anthropic is a major development in the highly competitive artificial intelligence industry, signaling Anthropic's aggressive pursuit of elite talent in its rivalry with companies like OpenAI, Google, and Meta. His expertise in deep learning, computer vision, and large language models is expected to significantly bolster Anthropic's research capabilities. The focus on using AI to accelerate its own pre-training research could lead to groundbreaking advancements in autonomous AI development and further cement Anthropic's position as a leader in the field.

Geographic Location

• Virtual/Online (Andrej Karpathy's announcement on X)