Google plans to move its roughly 90-person AI responsibility team from Google DeepMind into the company’s global affairs division, with the reorganization expected to take effect in early September, according to The Wall Street Journal.
Google’s global affairs unit handles the company’s lobbying and public policy work and also manages group-level trust and safety teams. The report said the change is part of Google’s broader integration of DeepMind into the wider company structure.
Leadership and support functions are also being reshaped
Before this move, DeepMind co-founder and chief executive Demis Hassabis had already shifted into a chairman role focused on external affairs and broader AI issues. His successor, Koray Kavukcuoglu, is set to run DeepMind as a Google senior vice president rather than as chief executive.
Human resources and policy teams are also being transferred from DeepMind into Google’s central departments.
The team works on model safety and user impact research
The AI responsibility team tests Google’s AI models for possible chemical, biological, radiological, and nuclear risks. It also studies how users interact with AI and the psychological effects of chatbots.
Some staff worry about reduced access to Gemini work
Some employees worry that moving the team out of DeepMind could reduce its access to and familiarity with Gemini model development. In their view, that could weaken the team’s ability to carry out independent research and spot risks in new models.
Google said centralizing the AI responsibility team would strengthen its influence over model and product safety work.
The report also said some employees had applied to transfer into research teams that will remain inside DeepMind, but those requests were denied. Management also acknowledged that the restructuring could lead some staff to leave for other AI labs.
The changes come as several senior DeepMind researchers have been leaving the company.