Karpathy Returns to Frontier LLM Work at Anthropic
*Andrej Karpathy has joined Anthropic to resume hands-on research on large language models after founding an education startup last year.*
The announcement
Karpathy posted the news on X on May 19. “Personal update: I’ve joined Anthropic,” he wrote. He added that he expects the next few years at the frontier of LLMs to prove especially formative and that he is eager to return to R&D. He also noted he remains committed to education and intends to pick that work up again later.
The post quickly reached the front page of Hacker News, where it drew 681 points and 264 comments within hours.
Career path
Karpathy co-founded OpenAI in 2015. He later served as director of AI at Tesla from 2017 to 2022, reporting directly to Elon Musk. He rejoined OpenAI in 2023, left the company again in 2024, and started Eureka Labs, an AI-focused education venture. In February 2025 he coined the term “vibe coding.”
His move to Anthropic therefore marks the third major lab switch in less than a decade and the second return to a frontier research role after a period outside it.
Immediate implications
Anthropic gains a researcher with direct experience at both OpenAI and Tesla’s autonomy stack. Karpathy gains an environment already shipping production models and a stated focus on scalable oversight and interpretability. The timing aligns with rapid iteration on next-generation LLMs at multiple labs.
No details have surfaced yet on his specific team or project assignments. Karpathy’s statement frames the decision as a return to core research rather than a shift into product or policy work.
Why it matters
Frontier labs continue to absorb the highest-profile talent even as the cost of training runs climbs and the technical questions grow more specialized. Karpathy’s record shows he can translate research into both widely used models and accessible teaching material. His presence at Anthropic strengthens one lab’s bench while removing him from the independent education track he started only last year. The field’s concentration of expertise therefore tightens further.
The next visible signal will be whatever Karpathy chooses to publish or ship from inside the company.
---
Sources:
{
"excerpt": "Andrej Karpathy has joined Anthropic to resume direct work on frontier large language models after founding an AI education startup last year.",
"suggestedSection": "ai",
"suggestedTags": ["andrej-karpathy", "anthropic", "llm-research"],
"imagePrompt": "Abstract layered diagrams of interconnected nodes hover above a sparse wooden desk holding only a single notebook and a dim desk lamp. Soft shadows stretch across matte surfaces, suggesting a quiet return to focused technical work. muted color palette, cinematic lighting, 16:9."
}
No comments yet