THE CLAIM. Google is relocating its roughly 90-person AI responsibility team, which evaluates frontier models for chemical, biological, radiological, and nuclear risk, out of DeepMind and into Google's global affairs division, the unit that handles lobbying and public policy, effective September 1, 2026. Leadership says the team's mandate and independence are unchanged. THE CHECK: confirmed by three independent outlets citing the same internal communication from VP Helen King, who says the team keeps its computing access, resources, and headcount. But the same reporting quotes unnamed employees and executives worried the move will cut their day-to-day access to the researchers actually building Gemini. THE TWIST: this is not the whole safety org moving, model behavior, privacy, and security functions explicitly stay inside DeepMind, so the headline 'Google guts AI safety' overstates it, while Google's 'nothing changes' undersells that moving a frontier-risk evaluation team into a policy and lobbying division is a real structural change in who that team reports to and sits near.