Google DeepMind, the company’s artificial intelligence development division, carried out a significant reallocation of its staff. Approximately 90 researchers, previously focused on AI security, were transferred to the Global Affairs sector, where they will manage general policies involving Google’s various services.
These experts are crucial in assessing potential biological, chemical, radiological and nuclear threats linked to artificial intelligence models. The presence of a team dedicated to AI security is essential to identify flaws in system responses, especially because they can delve into complex topics, allowing potential malicious uses to be tested.
Information about these internal changes was revealed through Google’s corporate emails. This move comes shortly after Demis Hassabis stepped down as CEO of DeepMind to assume the chairmanship of the board and the role of chief scientist at Alphabet, Google’s parent company.
The role of security in artificial intelligence in the current technological scenario
Rapid advances in the technology sector have reinforced the importance of professionals with knowledge in general sciences for managing the development of artificial intelligence. Competitors such as OpenAI and Anthropic have already expressed great concern about the evolution of their own models, highlighting the ability to delve deeper into scientific subjects and promising to incorporate security measures when dealing with sensitive topics.
Google announced that the transfer of these professionals aims to bring their activities closer to the product area, including the search tool, for example. However, among the employees themselves, there is concern that this change could limit the team’s access to the processes for creating new versions of Gemini, which could make it difficult to detect potential flaws in the system.
In an electronic message sent to employees, Helen King, vice president of DeepMind responsible for AI security, assured that the team will maintain full access to the innovations under development.
Other recent transformations at Google DeepMind
The migration of researchers should begin at the beginning of September, representing yet another change promoted by Google in its artificial intelligence division. In a previous move, in early August, DeepMind’s leadership was changed with the departure of Demis Hassabis as CEO. He was replaced by Koray Kavukcuoglu, while Jeff Dean became chief scientist. Furthermore, the Alphafold project, which earned Hassabis the Nobel Prize in Chemistry, was discontinued.
In addition to Dean, who had 27 years of experience at Google and was involved throughout the company’s AI development journey, other long-time experts also left the company. Many chose to found new startups or join other artificial intelligence research teams. Among them are Noam Shazeer and John Jumper, who went, respectively, to OpenAI and Anthropic. There is a fear in the market that recent changes to DeepMind could encourage more professionals to seek similar opportunities.
