AI
Jeff Dean's exit signals ethical rifts at Google as AI race accelerates
The departure of a longtime Google veteran and other leadership reshuffles appear driven by pressure to speed up products—and growing internal conflicts over military AI contracts.
The Verge
Google announced sweeping changes to its AI leadership this week, but behind the unified messaging lies a more turbulent reality: tensions between research ambitions and product velocity, compounded by friction over military partnerships.
The reshuffling#
The restructuring involves several significant moves. DeepMind leader Demis Hassabis is stepping back from day-to-day operations to focus on artificial general intelligence research and the chair role, with CTO Koray Kavukcuoglu taking over leadership. More notably, Jeff Dean—a 27-year veteran who was Google's 30th employee—is departing along with three other senior AI researchers (Sanjay Ghemawat, Quoc Le, and Oriol Vinyals) to launch Discovery Loop, a startup focused on automating scientific experiments in drug discovery, clean energy, and materials design.
Speed versus research#
Industry observers read the changes as a victory for product-focused momentum over Hassabis's longer-term research interests. Google executives acknowledged the company needs to move faster: CEO Sundar Pichai stated the organization must "continue to move fast and with clear purpose," while acknowledging competition from OpenAI and Anthropic remains fierce.
Sources within Google described unprecedented internal competition around AI products and pressure to hire contributors who can accelerate launches. One former Google employee told The Verge there is "far more internal competition at Google than I've seen in the last decade primarily around the AI product space."
The ethics question#
Beneath the structural reshuffling lies a deeper fault line: Google's willingness to work with the Department of Defense. The company recently agreed to allow Pentagon use of its AI for "any lawful use"—a move that notably filled a gap left when Anthropic maintained red lines prohibiting domestic mass surveillance and lethal autonomous weapons systems.
Dean had previously aligned with Anthropic's position. In February, he posted on X that "mass surveillance violates the Fourth Amendment," and in 2018 he and Hassabis signed a public pledge against delegating lethal decisions to machines. Yet Google's Pentagon deal stands at odds with those positions.
The AI industry watchdog The Midas Project's founder Tyler Johnston told The Verge the defense contract "has already caused staff departures and a unionization effort at DeepMind," and suggested Dean and Hassabis may have personally opposed the move. DeepMind was by Google with an explicit agreement restricting military use.

