AI researchers continue to leave Google for its rivals

A wave of senior AI researchers is departing Google for rival labs Anthropic and OpenAI, according to a report. The exits include a Nobel laureate and key contributors to Google's Gemini model, intensifying a talent battle as competitors prepare to go public. Jonas Adler and Alexander Pritzel, who played key roles in developing Google's Gemini model, are leaving for Anthropic, Bloomberg reported [1]. Gemini is a family of multimodal large language models developed by Google DeepMind and announced in December 2023 [11]. Their departures follow two other high-profile exits from Google in the same month [1]. Noam Shazeer, a researcher who had been at Google since 2000, announced he was leaving for OpenAI [1]. Shazeer spent three years building the chatbot startup Character.AI, which Google effectively acquihired for $2.7 billion, in part to bring him back to work on Gemini [1]. Days later, Google DeepMind Director John Jumper said he was leaving for Anthropic [1]. Jumper won the 2024 Nobel Prize in Chemistry alongside DeepMind CEO Demis Hassabis for his work on AlphaFold, a system that predicts 3D protein structures from amino acid sequences [1]. DeepMind had made significant advances in protein folding with AlphaFold in 2020, and by July 2022 had released over 200 million predicted protein structures [3]. Anthropic, founded in 2021 by former OpenAI members, has developed the Claude series of large language models and maintains a focus on AI safety [2]. The company is privately held but plans to go public, with an estimated valuation of $965 billion in May 2026 [2]. OpenAI, which released ChatGPT in November 2022, conducted a $6.6 billion share sale in October 2025 that valued the company at $500 billion [5]. Throughout 2024, roughly half of then-employed AI safety researchers left OpenAI, citing the company's deprioritization of safety goals [5]. A 2025 analysis of 1,178 safety and reliability papers found that corporate AI research increasingly concentrates on pre-deployment areas such as model alignment and testing, while attention to deployment-stage issues like model bias has waned [8]. The same year, four frontier AI labs — Anthropic, Google DeepMind, OpenAI, and xAI — reported ARC-AGI benchmark performance in public model cards, establishing it as an industry standard for AI reasoning [6]. As Anthropic and OpenAI prepare to go public, the promise of equity could continue to draw top AI talent from established labs [1].