Google Reorganizes AI Safety Unit and Hires Barret Zoph

Google is integrating its DeepMind research lab into its standard corporate structure, moving away from the semiautonomous model the lab has maintained since 2014. As part of this reorganization, the company is transferring its 90-person AI responsibility unit from DeepMind to the global affairs division. This unit specializes in human-AI interaction, the psychological impacts of chatbots, and risks involving chemical, biological, radiological, and nuclear materials.

The restructuring follows an August 5 announcement that moved DeepMind CEO Demis Hassabis into a chairman role. To bolster its research leadership, Google DeepMind hired Barret Zoph as Vice President of Research. Zoph previously served as Vice President of Research at OpenAI and co-founded the startup Thinking Machines Lab.

While a company spokesperson stated the move strengthens the ability to inform safety for models and products, some employees expressed concerns that the new reporting structure will reduce insight into the latest AI models and hamper independent research.