Google shifts its AI responsibility unit from DeepMind to global affairs division

Starting in September, Google plans to relocate its AI responsibility team—about 90 staff members—from the DeepMind subsidiary to the global affairs organization that handles lobbying and public policy. Helen King, a vice president at DeepMind, explained the change as part of a larger transition toward a more typical product-area structure within Google. While the team will retain access to DeepMind’s resources and headcount, some members fear reduced proximity to DeepMind researchers could hinder independent safety research, especially concerning Gemini AI models.

The unit conducts work on chemical, biological, radiological, and nuclear risk testing as well as studies on human-AI interaction. Google says the reorganization aims to bring AI safety efforts together across the company to better inform model and product safety. Demis remains supportive of responsible-AI initiatives, according to internal communications.