OpenAI Co-Founder Andrej Karpathy Just Joined Anthropic. Why This Is the Biggest Talent Move of 2026

Andrej Karpathy announced on X on May 19, 2026 that he has joined Anthropic. The post landed early Tuesday morning and the AI industry has not been able to talk about much else since. Karpathy is not just any researcher. He is one of the original founding members of OpenAI, former senior director of AI at Tesla where he led Autopilot and Full Self-Driving, and one of the most influential figures in modern machine learning education through his Neural Networks: Zero to Hero course and YouTube lectures.

I’ve joined Anthropic. 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.

The hire is one of the most significant talent moves in AI this year. It also fits a pattern that has been building for two years: senior OpenAI talent leaving the company that built ChatGPT, often for Anthropic, often for similar reasons. Read alongside the other Anthropic news this week, the Karpathy signing is more than a hire. It is a statement about where the center of frontier AI search and AEO research is moving.

Who Is Andrej Karpathy and Why Does This Matter?

In one line: Karpathy is one of the few researchers who can bridge LLM theory and large-scale training practice, which is exactly what Anthropic just hired him to do. Karpathy studied under Fei-Fei Li at Stanford and joined OpenAI as one of its earliest staffers. He left in 2017 to lead AI at Tesla, where he ran the Autopilot and Full Self-Driving computer vision teams until 2022. He returned to OpenAI in 2023, then left again in 2024 to start Eureka Labs, an AI-integrated education startup. He is also widely credited with coining the term “vibe coding” in early 2025, which has since become shorthand for AI-assisted software development across the industry. What makes Karpathy specifically valuable to Anthropic is the rare combination of deep theoretical understanding of large language models with hands-on experience operating massive training runs at scale. Most researchers specialize in one or the other. Karpathy has done both, repeatedly, at companies where the cost of being wrong about either is measured in hundreds of millions of dollars.

What Will Karpathy Actually Be Working On at Anthropic?

In one line: Pre-training, plus a new team focused on using Claude to accelerate AI research itself. Karpathy started this week under Nick Joseph, who leads Anthropic’s pre-training team. According to TechCrunch’s reporting, pre-training is responsible for the large-scale training runs that give Claude its core knowledge and capabilities. It is the most expensive and compute-intensive phase of building a frontier model, and the work that most directly determines what a model can do before any fine-tuning or alignment work begins. Beyond joining the existing team, Karpathy will help launch a new team specifically focused on using Claude to accelerate pre-training research. That framing matters. It signals that Anthropic is making an explicit bet that AI-assisted research, rather than pure compute scaling, is how it stays competitive with OpenAI and Google in the next phase of frontier model development. If the lab that figures out how to use current AI models to build better future AI models pulls ahead, the compounding effect could be significant.

How Many OpenAI Co-Founders and Senior Leaders Have Left?

In one line: The OpenAI exodus has been steady for two years, with senior research talent consistently moving toward safety-focused labs. Karpathy is the latest in a notable series of senior OpenAI departures. Anthropic itself was founded in 2021 by Dario and Daniela Amodei, both former OpenAI executives. With John Schulman in 2024 and now Karpathy in 2026, the company has effectively absorbed multiple founding-era OpenAI researchers. Reuters reported that the pattern is not random. OpenAI has shed senior research leadership consistently since 2024, and a disproportionate share of those departures has landed at Anthropic specifically rather than at Google DeepMind, Meta AI, or independent ventures.

Why This Week Specifically Matters for Anthropic

In one line: Three structural wins for Anthropic landed inside seven days. Read in isolation, the Karpathy hire is a major talent win. Read in context, it lands inside a week where three separate structural advantages for Anthropic compounded. None of these three are causally linked, but they are correlated by something deeper: a structural shift in where the most respected research talent, the cleanest user experience, and the fastest-growing AI referrals are all moving. The same direction. At the same time.

What Does This Mean for AI Search and AEO?

In one line: If Anthropic’s research advantage compounds, Claude’s capabilities will pull further ahead and the platforms it powers will get more important for AEO. The direct AEO implication is not that Karpathy alone changes citation behavior overnight. It is that the model quality powering Claude is now being shaped by one of the few researchers in the world who can move pre-training capability meaningfully forward. Pre-training improvements affect everything downstream: reasoning quality, factual recall, citation behavior, context retention. If Claude’s reasoning and citation quality continue to pull ahead of ChatGPT and Gemini, the brands optimizing for Claude visibility through AEO now will be early to a surface that is about to matter more. The secondary implication is about Claude’s growth trajectory. Major hires drive media attention, which drives platform trials, which drives long-term share. Claude’s referral share already tripled in the first four months of 2026. The Karpathy news cycle alone will likely accelerate that further. For AEO practitioners who have been treating Claude as a secondary platform, this is the moment that thesis stops being defensible. For a deeper look at how citation behavior differs across AI platforms, the what is AEO breakdown on Prompt Insider explains the structural differences between ChatGPT, Perplexity, Claude, and Gemini in detail.

The Other Hire That Almost Got Missed: Chris Rohlf

In one line: Anthropic also added a 20-year cybersecurity veteran to its frontier red team the same day. Lost in the Karpathy news cycle: Anthropic also announced on May 19 that Chris Rohlf is joining the company’s frontier red team, which stress-tests advanced AI models against severe threats. Rohlf has more than 20 years of cybersecurity experience, including six years at Meta and an earlier stint at Yahoo’s well-respected cybersecurity team known as The Paranoids. He was also a fellow at Georgetown’s Center for Security and Emerging Technology working on the CyberAI project. “We have a real opportunity in front of us to dramatically improve cyber security with AI,” Rohlf posted on X. “I can’t think of a better company or team to join at this critical moment in time.” The dual hire signals that Anthropic is staffing up both the offensive capability side (Karpathy on pre-training) and the defensive safety side (Rohlf on red teaming) simultaneously, which is consistent with the company’s stated approach of building frontier capabilities and frontier safety work in parallel rather than sequentially.

Frequently Asked Questions

When did Andrej Karpathy join Anthropic?

Karpathy announced on X on May 19, 2026 that he has joined Anthropic and started this week on the company’s pre-training team under team lead Nick Joseph.

What will Karpathy be doing at Anthropic?

Karpathy is working on pre-training, the large-scale training phase responsible for Claude’s core knowledge and capabilities. He will also help launch a new team focused specifically on using Claude itself to accelerate pre-training research, which signals Anthropic’s bet that AI-assisted research is how it stays competitive at the frontier.

Is Karpathy an OpenAI co-founder?

Yes. Karpathy was one of the founding members of OpenAI and one of its earliest staffers. He left in 2017 to lead AI at Tesla, returned to OpenAI in 2023, then left again in 2024 to start Eureka Labs before joining Anthropic in May 2026.

How many OpenAI co-founders are now at Anthropic?

Two confirmed OpenAI co-founders are now at Anthropic: Andrej Karpathy (joined May 2026) and John Schulman (joined 2024). Anthropic itself was co-founded in 2021 by Dario and Daniela Amodei, both former OpenAI executives, so the founding leadership of the company also traces back to OpenAI.

What is Karpathy known for outside of OpenAI?

Karpathy led AI at Tesla from 2017 to 2022, where he ran the Autopilot and Full Self-Driving computer vision teams. He is also widely known for his Neural Networks: Zero to Hero course and YouTube lectures, which have become standard educational resources for new machine learning researchers. He is credited with coining the term “vibe coding” in early 2025, and founded Eureka Labs, an AI-integrated education startup, in 2024.

Did Anthropic announce any other hires the same day?

Yes. Anthropic announced on May 19 that Chris Rohlf is joining the company’s frontier red team, which stress-tests advanced AI models against severe threats. Rohlf has more than 20 years of cybersecurity experience, including roles at Meta, Yahoo, and Georgetown’s Center for Security and Emerging Technology.

What does the Karpathy hire mean for AEO?

The direct AEO implication is that Claude’s underlying model quality is now being shaped by one of the few researchers in the world capable of meaningfully advancing pre-training. If Claude’s reasoning and citation behavior continue to improve faster than competitors, the brands optimizing for Claude visibility through AEO now will be early to a surface that is increasingly important. The hire also amplifies Claude’s media attention and growth momentum at a moment when ChatGPT referral share is declining.