Google DeepMind investigates risks of billions of AI agents interacting online
2026-06-11 · MIT Tech Review AI · Original
Google DeepMind is funding crucial research into the safety risks associated with large-scale AI agent interactions. As autonomous agents capable of executing tasks without human intervention become increasingly common, concerns are growing about their ability to follow instructions from other AI systems. Rohin Shah, leading AGI safety efforts at DeepMind, warns that the complex web of millions of autonomous bots could create unforeseen consequences, making understanding these dynamics a top priority for the company.