Google shakes up AI teams structure

By the Editorial Team · Friday, August 28, 2026

Google is reorganizing its AI responsibility team by moving it out of DeepMind and into the company’s global affairs division, aligning it more closely with public policy and lobbying efforts. This change aims to unify AI safety initiatives across Google, though some team members worry it may limit their independent research and direct access to DeepMind’s core AI development projects.

Google is restructuring its artificial intelligence (AI) teams by moving the group focused on AI risks and societal impact out of Google DeepMind, the company’s research lab responsible for advanced AI model development. The approximately 90-person “AI responsibility” unit will be integrated into Google’s global affairs organization starting next week. This division handles the company’s lobbying and public policy efforts.

The reorganization follows a leadership change in DeepMind, with co-founder and CEO Demis Hassabis stepping down to take on a new chairman role. Koray Kavukcuoglu has succeeded Hassabis as head of DeepMind, assuming the title of senior vice president, rather than CEO. The shift is part of a broader effort to more closely align DeepMind with Google’s overall operations. DeepMind, acquired by Google in 2014, has traditionally operated with significant autonomy.

Helen King, a Google DeepMind vice president who leads the responsibility team, explained in an internal email that the move brings the AI responsibility group into a typical product area setup similar to other Google divisions such as Google Search. Other support functions like human resources and policy are also transitioning from DeepMind into central Google.

Some employees within the responsibility unit have expressed concerns about the impact of the reassignment on their ability to conduct independent research. They worry that distancing the team from DeepMind’s core AI developers—who work on projects like Google’s Gemini AI models—could reduce access to vital information and hinder the identification of pertinent research questions. The team conducts work on a range of issues including testing AI models for chemical, biological, radiological, and nuclear risks, as well as studying human interactions with AI systems and psychological effects of chatbots.

A Google spokesperson stated the move is intended to consolidate AI safety and responsibility efforts across the company, enhancing the teams’ ability to improve safety measures for Google’s products and models. Notably, other AI safety groups focused on model behavior, privacy, and security will remain within DeepMind. The global affairs organization also houses Google’s central trust and safety team.

King has sought to reassure staff that the team’s core mission and day-to-day work will not change. She emphasized that the group will continue to focus on emerging and frontier topics in responsible AI and maintain its commitment to external engagement through publications and events. The responsibility team will keep its current computing resources and staffing levels, and maintain access to DeepMind’s infrastructure. King added that Hassabis remains committed to responsible AI and plans to stay involved in related efforts.

This article was reported by synthesizing coverage from multiple independent news outlets to present a balanced, unbiased account. Read about our editorial standards. Spotted an error? Let us know.