Andrej Karpathy joins Anthropic for LLM pre-training research

Andrej Karpathy, a prominent figure in artificial intelligence and a founding member of OpenAI, has announced he is joining Anthropic. The news came via a post on his X account on May 19, 2026, where Karpathy expressed excitement about returning to hands-on research and development in large language models (LLMs). He specifically noted that the next few years will be 'especially formative' for the field. Karpathy will be working under team lead Nick Joseph, focusing on pre-training—a critical phase where models learn from massive datasets before fine-tuning. His move marks a significant talent acquisition for Anthropic as it competes with OpenAI and other labs in the rapidly evolving AI landscape.

Karpathy's background includes serving as Director of AI at Tesla, where he led Autopilot vision, and playing a key role in OpenAI's early breakthroughs. His decision to join Anthropic, rather than return to OpenAI or start another venture, underscores the company's growing influence. The focus on pre-training suggests Karpathy will work on foundational model architecture and scaling strategies, areas where his expertise is highly valued. This hire could accelerate Anthropic's efforts to build safer, more capable LLMs, potentially challenging the current market leaders.