On June 11, 2026, Google DeepMind, together with Schmidt Sciences, the Cooperative AI Foundation, the Advanced Research and Invention Agency (ARIA), and supported by Google.org, announced a new technical research funding call of up to $10 million for researchers worldwide. The initiative targets the safety challenges posed by the upcoming era of millions of interacting AI agents built by different organizations.
Why Multi-Agent Safety Matters
As AI technology scales, autonomous agents will communicate, negotiate, and transact across digital environments. When large groups of AI agents interact, new collective behaviors and capabilities can emerge suddenly. Current safety evaluations analyze models in isolation, but interacting agents can produce complex emergent behaviors that are difficult to anticipate. Understanding these system-wide behaviors is critical to prevent unpredictable economic activity or new security challenges.
Scaling Research
Although foundational frameworks exist, the rapid evolution of multi-agent systems requires immediate large-scale expansion. Google DeepMind's 2025 research established a framework for understanding interactions, and recent work on AI Agent Traps explores vulnerabilities in adversarial environments. This funding call aims to accelerate progress by supporting a global network of independent researchers, ensuring safety standards are transparent and robust.
Priority Research Areas
- Sandboxes and testbeds: Building realistic, reproducible environments to evaluate and compare multi-agent safety, including virtual marketplaces, simulated ecosystems, and multi-organisation workflows.
- The science of agent networks: Understanding safety-relevant properties of interacting agent populations, such as emergent capabilities, network failures, volatility, and detection of dangerous population-level properties.
- Strengthening agent infrastructure: Stress-testing protocols for identity, reputation, and commitment that enable secure cross-platform agent interactions.
- Oversight and control: Developing methods to monitor deployed agent populations and mitigate collective harms at scale.
How to Participate
Researchers are invited to submit proposals. The deadline to apply is August 8, 2026, with awardees expected to be announced in Autumn 2026. Details and the application portal are available on the Google DeepMind website.
Images
