Top AI researcher Andrew Tulloch ditches Meta to work on training and inference at Anthropic

Prominent AI researcher Andrew Tulloch is making a significant career shift, departing from Meta after a brief tenure to join Anthropic. Starting next week, Tulloch will focus on enhancing the training and performance of Anthropic’s AI models within their inference team. Tulloch's impressive background, which includes an 11-year stint at Meta and a role at OpenAI where he contributed to the development of the GPT-4 and GPT-4.5 models, has made him one of the most sought-after talents in the competitive AI landscape. Notably, he previously co-founded Thinking Machines Lab in 2025, illustrating his entrepreneurial spirit. Despite being offered a staggering pay package of up to $1.5 billion from Meta's CEO Mark Zuckerberg, Tulloch chose to join the social media giant only last year, before now deciding to switch to Anthropic. This move comes shortly after Meta's launch of Muse, an AI agent designed to assist users autonomously. The ongoing battle among tech giants like Anthropic, Meta, OpenAI, and Google for top AI researchers has intensified since the launch of ChatGPT, providing professionals like Tulloch with substantial leverage in negotiations. As Anthropic gears up for its initial public offering, Tulloch's expertise could provide a stabilizing force amid a tumultuous industry landscape. His transition follows the recent resignation of Jacob Coxon from Anthropic, who expressed concerns about the potential risks of AI technologies. Both Tulloch and Meta have not publicly commented on this transition, but it highlights the ongoing shifts in the AI sector as companies vie for innovation and talent amid rising scrutiny over the implications of their technologies.