Back to feed
News Story
AI前线
1 sources

Karpathy Rumored to Leave Anthropic After Just Two Months

AI luminary Andrej Karpathy, who joined Anthropic only two months ago, has sparked departure rumors after removing Anthropic references from his social media bio. Unconfirmed reports claim he submitted a resignation, but neither he nor Anthropic has confirmed. His hiring in May was seen as a major talent win for the company.

SynthePulse Insight · AI deep reading

Karpathy's 68 Days: From Joining Anthropic to Deleting His Bio, a Public Image Tug-of-War Over Open vs. Closed Source

Version 1 · 2 sources

Andrej Karpathy modified his X bio, removing Anthropic info, sparking departure rumors. Behind the event lies the ongoing tension between his identity as an open-source educator and a closed-source company employee, as well as Anthropic's absence from an open-weight open letter.

  • Andrej Karpathy joined Anthropic on May 19, 2026, as a technical staff member, tasked with building a research team to accelerate pretraining research using Claude.
  • Around July 26, 2026, Karpathy modified his X bio, deleting Anthropic-related information, triggering departure rumors.
  • Some netizens claimed Karpathy had submitted a resignation letter, but the news is unverified; Karpathy himself later denied leaving, calling it 'fake news.'
  • During Karpathy's 68 days at Anthropic, he experienced the troubled launch of Claude Fable 5, controversy over the Claude Tag product promotion, and Anthropic's failure to sign an open-weight open letter.
  • Karpathy was previously known for open-source education projects (e.g., nanoGPT, CS231n), and his move to closed-source Anthropic sparked public debate about his role change.
Open section navigationBio Modification Sparks Departure Speculation

Bio Modification Sparks Departure Speculation

Around July 26, 2026, netizens noticed that Andrej Karpathy had modified his X bio, removing work information related to Anthropic. This detail quickly sparked external speculation: had Karpathy left Anthropic? A blogger shared unverified news that Karpathy had submitted a resignation letter and thanked Anthropic for the opportunity, but the claim lacked reliable sources or official confirmation.

Subsequently, Tibo, head of OpenAI Codex, seemingly stepped in to clarify the rumors. Karpathy himself also responded on X: 'Twitter is spreading this weird fake news. No, I haven't resigned.' He added that if he were to resign, he wouldn't do it by changing his bio. Karpathy's Wikipedia page still shows 'Anthropic (2026–present).'

Controversial Events Within 68 Days

Karpathy officially announced his joining Anthropic on May 19, 2026, entering the pretraining team led by Nick Joseph, tasked with using Claude to accelerate pretraining research. However, the 68 days after joining were not peaceful. On June 9, Anthropic released Claude Fable 5, but it was quickly disabled for various reasons, with permissions repeatedly adjusted—a rocky start.

On June 23, Anthropic launched Claude Tag, integrating Claude into Slack. Karpathy highly praised the product, calling it the 'third major redesign' of large model interaction. However, some netizens felt the praise was excessive and criticized Karpathy for transforming from an educator into a company man, endorsing closed-source products.

More pointed criticism came from the open-source community: Karpathy had taught people to implement neural networks from scratch through projects like nanoGPT and micrograd, yet now he was defending closed-source products that do not disclose model weights or training data. This role change made many netizens nostalgic for his image as a neutral educator.

Open-Weight Open Letter and Anthropic's Absence

Around July 24, 2026, Jensen Huang posted his first X post, sharing a joint open letter supporting open-weight AI models. The letter was initially signed by 25 companies including NVIDIA, Microsoft, and Meta, but notably absent were OpenAI, Google, and Anthropic. Subsequently, OpenAI and Google joined, but Anthropic remained absent.

Netizens linked Karpathy's bio deletion to Anthropic's failure to sign the open letter, speculating that he might have hidden his work information due to dissatisfaction with his employer's stance. Some netizens joked, 'He might just be ashamed of his employer.' However, this causal relationship has not been confirmed. In the controversy, an Anthropic employee retorted, 'Why doesn't NVIDIA open-source CUDA?' Andrew Ng responded that one cannot block others from open-sourcing while being closed-source themselves.

Public Image Tug-of-War: Open-Source Educator vs. Closed-Source Company Employee

Karpathy's uniqueness lies in his multiple identities: OpenAI founding member, former Tesla AI director, Stanford CS231n instructor, and educator who popularized AI technology through minimalist open-source projects like nanoGPT and llm.c. He is seen as a representative of open technology and inclusive education.

Therefore, after joining the highly closed-source Anthropic, his praise for products, silence on controversies, and even bio modifications have been repeatedly interpreted by the public. The core of this storm is not whether Karpathy left, but the public's ongoing attention and expectation gap regarding his role change.

Karpathy himself denied leaving, but the incident highlights the tension between open-source and closed-source paths in the AI field, as well as the public relations risks when personal brand conflicts with company stance.

Credibility boundary

This article is based on reports from Qubit and Machine Heart. The departure rumors primarily stem from Karpathy's X bio modification, but Karpathy himself has denied them. Netizen leaks and screenshots are unverified. Other events (such as the Fable 5 launch, Claude Tag release, and open letter signing) have clear timelines and sources, but some causal links (e.g., bio modification and open letter absence) are speculative.

Insight takeaway

Karpathy's bio modification incident is a microcosm of the conflict between open-source and closed-source paths in AI, reflecting the public's high expectations for consistency in tech leaders' roles. The incident itself has been clarified, but discussions around Karpathy's identity will continue.

Primary report

AI前线

Primary source