After less than a year at Meta, prominent AI researcher Andrew Tulloch is moving to Anthropic, where he will join the inference and performance team. Tulloch will focus on optimizing the training of Anthropic’s AI models and improving their responses to users.
Tulloch’s departure comes after he turned down a significant pay offer from Meta CEO Mark Zuckerberg, which could have amounted to $1.5 billion. He joined Meta in October 2023, having previously worked at OpenAI, where he contributed to the development of GPT-4o, GPT-4.5, and o3 models. Before OpenAI, Tulloch cofounded Thinking Machines Lab in 2025.
Tulloch’s extensive experience in AI research is expected to be a valuable asset to Anthropic, which is preparing for its initial public offering, competing with rivals, and addressing concerns about the potential adverse effects of AI.
An Anthropic spokesperson confirmed Tulloch will start his new role next week. “We are excited to have Andrew join our team,” the spokesperson said. “His expertise will be crucial as we continue to optimize our AI models and enhance user interactions.”
Meta, where Tulloch had worked for 11 years, recently launched Muse, an AI agent designed to autonomously complete tasks for users. Anthropic, meanwhile, is facing intense competition for AI talent, which has given experienced researchers like Tulloch significant leverage.
Semafor reported that Tulloch’s departure from Meta was imminent. The move comes as Anthropic is navigating a challenging period, with concerns about the potential risks associated with advanced AI models.
In a separate development, Anthropic employee Jacob Coxon recently resigned and expressed serious concerns about the potential risks of AI, stating, “the people building AI earnestly believe that it could kill us all by the end of the decade.”
For now, Anthropic has not commented on Tulloch’s hiring, and Meta has not responded to requests for comment.
Tulloch’s departure highlights the ongoing talent wars in the AI industry, where companies are competing fiercely for top researchers to fuel their advancements in AI technology.
Thinking about building an AI product?
Get in Touch