Andrej Karpathy, one of the most respected names in AI research, has made another career move that’s sending ripples through the industry. The OpenAI co-founder and former Tesla AI chief has joined Anthropic to work on pre-training research.
“I’ve joined Anthropic,” Karpathy posted on X Tuesday. “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 started this week at Anthropic, where he’s working on pre-training under team lead Nick Joseph. This is significant because pre-training handles the massive, expensive training runs that give Claude its core knowledge and capabilities. It’s one of the most compute-intensive and costly phases of building advanced AI models.
An Anthropic spokesperson told TechCrunch that Karpathy will start a team focused on using Claude to speed up pre-training research. This move signals Anthropic’s belief that AI-assisted research, rather than just throwing more compute power at problems, is how it will stay competitive with OpenAI and Google.
The hire makes strategic sense. Karpathy is one of the few researchers who can bridge the gap between LLM theory and large-scale training practice. His track record speaks for itself:
- Co-founded OpenAI and worked on deep learning and computer vision until 2017
- Led Tesla’s Full Self-Driving and Autopilot programs from 2017 to 2022
- Returned to OpenAI for one year before leaving again in 2024
- Started Eureka Labs, an AI education startup, though updates have been sparse
Karpathy’s move comes at a critical time in the AI race. As training costs soar and compute becomes increasingly expensive, companies are looking for smarter ways to build better models. Using AI to improve AI development itself could be the next breakthrough.
“I remain deeply passionate about education and plan to resume my work on it in time,” Karpathy said, suggesting his education-focused projects aren’t completely off the table. He’s known for his popular Neural Networks: Zero to Hero course and YouTube channel where he teaches AI concepts.
Anthropic also brought on cybersecurity veteran Chris Rohlf to its frontier red team, which tests advanced AI models against serious threats. Rohlf has over 20 years of experience, including time at Yahoo’s cybersecurity team “The Paranoids” and six years at Meta. He was also a fellow at Georgetown’s Center for Security and Emerging Technology.
“We have a real opportunity in front of us to dramatically improve cyber security with AI,” Rohlf said on X. “I can’t think of a better company or team to join at this critical moment in time.”
These hires show Anthropic is doubling down on both advancing AI capabilities and ensuring safety. With competition heating up between major AI labs, having researchers of Karpathy’s caliber could give Anthropic a significant edge in developing the next generation of AI models.
