OpenAI co-founder Andrej Karpathy announced his exit just a day after it won a high-stakes court battle against Elon Musk. Karpathy, on Tuesday, confirmed that he has joined Anthropic as it sets its eyes on a $900 billion valuation. Karpathy, one of the eleven founder members of OpenAI, will now be working on AI-related research and development for Anthropic he said in his statement.
Karpathy has played a crucial role in building the fundamental technologies that shaped OpenAI into the AI powerhouse that it is today. His research contributed to OpenAI’s early generative models, AI training experiments, and merging natural language with image recognition that now allows ChatGPT to view an image and describe it in a sentence.
An AI veteran with an impressive track record
In June 2017, however, Karpathy was poached by Elon Musk to serve as Tesla’s Director of AI and Autopilot Vision where he worked till 2022.
Karpathy, did however, return to OpenAI between 2023 and 2024 wherein he worked on refining ChatGPT’s capabilities in generating ways to have AI models generate high-quality, synthetic text and data to train newer, smarter models. This was done to address the AI talent deficit in the human workforces at the time.
Updating about his pivot to Anthropic, he expressed excitement on returning to R&D.
“I think the next few years at the frontier of LLMs will be especially formative. I remain deeply passionate about education and plan to resume my work on it in time,” Karpathy noted.
Boris Cherny, the creator and head of the Claude Code at Anthropic welcomed Karpathy on the team and joined tech researchers and crypto firms like Jane Manchun Wong and Bitget in appreciating the update.
Exodus from OpenAI continues
Karpathy now joins a visibly elongating list of AI talent that is migrating out of OpenAI to join boutique AI startups or competing firms like Anthropic.
In recent times, a number of senior level OpenAI executives have taken up jobs in Anthropic. These include Andrea Vallone, who worked on model policy and safety research at OpenAI and Jan Leike, the former co-lead of OpenAI’s Superalignment project.
When Leike left OpenAI in 2024, he had expressed concerns regarding the company prioritizing commercial product launches over leading AI-centered safety researches.
Nicholas Joesph, one of the former foundational engineers at OpenAI, also joined Anthropic in 2021. On Tuesday, he welcomed Karpathy onboard revealing that the new joinee would be building a team focused on using Claude to accelerate pretraining research itself.
The company recently faced internal backlash for entering a controversial agreement with the U.S. Department of Defense (DoD) to deploy OpenAI’s advanced AI systems within the military’s classified networks. The deal involves OpenAI allowing the DoD to use its technologies for military purposes. Anthropic, on the other hand, had refused to compromise on its safety boundaries concerning military use and rejected the $200 million agreement and sued the U.S. government over threats to ban its technologies for wider usage.
OpenAI finally saw some respite amid a long-ongoing flurry of controversies this week after a federal jury tossed out Musk’s case against which had alleged that the AI giant had deviated from its original mission of being a non-profit organization to pivot to its new “for-profit” model. Musk now plans to file an appeal with the Ninth Circuit stating that, “creating a precedent to loot charities is incredibly destructive to charitable giving in America.”
