Google DeepMind is worried about what happens when millions of agents start to interact

Google DeepMind has allocated $10 million to support research on the risks associated with interactions among artificial intelligence agents. This funding aims to study the behavior of multi-agent systems and develop methods to prevent unsafe scenarios. Partners in this initiative include the Schmidt Science Foundation, ARIA, the Partnership on AI, and Google.org. This funding represents a significant step toward enhancing academic research in agent safety, an area that has previously lacked specialized studies.