Andrej Karpathy, a computer scientist and AI researcher best known as the co-founder of OpenAI and former director of AI at Tesla, announced today that he has joined Anthropic, the AI safety-focused company behind the Claude family of large language models.
Karpathy described the coming years at the frontier of LLMs as “especially formative” and said he was eager to return to research and development.
“I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D,” Karpathy said in his announcement. He also noted that he remains committed to education and intends to resume that work in the future.
Karpathy previously served as senior director of AI at Tesla, where he led the Autopilot computer vision team, and before that held a research role at OpenAI during its early years.
After leaving Tesla in 2022, he worked independently, producing widely followed educational content on neural networks and deep learning. That period made him one of the most recognized figures in AI without a corporate affiliation.
Anthropic was co-founded in 2021 by former OpenAI executives Dario and Daniela Amodei. The San Francisco-based company focuses on AI safety and alignment research. By 2025, funding rounds had made it one of the most capitalized private AI firms globally. Its Claude models have gained traction among enterprise customers and developers as an alternative to OpenAI’s GPT series and Google’s Gemini lineup.
The hiring strengthens Anthropic’s research team as competition among frontier AI labs intensifies. OpenAI, Google DeepMind, Meta AI, and xAI (Elon Musk’s AI venture) are all scaling their teams and compute infrastructure. Recruiting Karpathy signals that Anthropic intends to compete on technical capability alongside its safety work.
Apart from his roles at OpenAI and Tesla, Andrej Karpathy is also known for coining the term “vibe coding” in early 2025. He used the term on X to describe programming where developers specify intent in natural language and AI agents generate most of the code.
“Vibe coding” has since become a widely used label for AI-assisted development workflows, reflected in tools like Cursor, Replit, and similar platforms that emphasize rapid, prompt-driven software creation.
