Andrej Karpathy, a founding member of OpenAI and former director of AI at Tesla, announced on Tuesday, May 19, 2026, that he had joined Anthropic, the company behind Claude.
Nicholas Joseph, Anthropic’s pretraining lead, said Karpathy would build a team focused on using Claude to accelerate pretraining research itself. Karpathy said he was returning to research and development.
Using Claude for pretraining research
Anthropic describes pretraining as the large-scale model training that gives Claude its core knowledge and capabilities. Joseph’s announcement outlined a research objective without providing measured results or a timetable for changes to Claude.
Anthropic also makes Claude Code, a terminal-based coding tool. The company says it can read and edit files, write and run tests, and use command-line tools.
A career spanning AI research and education
The move takes Karpathy to OpenAI competitor Anthropic after two periods at OpenAI. He left OpenAI for Tesla in 2017, served as Tesla’s director of AI and led the computer vision team for Autopilot before leaving in 2022. He returned to OpenAI in 2023 and departed again in 2024.
In 2024, he founded Eureka Labs, an education venture introduced on July 16. Its launch announcement described an approach pairing teacher-designed course materials with AI teaching assistants to guide students through them.
Karpathy’s educational work also includes Neural Networks: Zero to Hero, a lecture series on building neural networks in code, and videos explaining large language models.
In his May 19 announcement, Karpathy said, “I remain deeply passionate about education and plan to resume my work on it in time.”
