DeepMind and partners commit up to $10M to multi-agent safety

Google DeepMind, Schmidt Sciences, the Cooperative AI Foundation, ARIA, and Google.org opened a research funding call of up to $10 million for multi-agent AI safety. The program targets risks that emerge when large populations of independently built agents communicate, negotiate, transact, and coordinate across networks. Priority areas include realistic sandboxes, the science of agent networks, secure identity and reputation infrastructure, and oversight methods for deployed agent populations. The initiative is high signal because most current evaluations study models individually, while future failures may emerge from interactions among many capable systems rather than from a single model in isolation.

Key details: June 11, 2026, Up to $10M research funding, Five partner organizations, Applications close August 8, 2026.

Continue swiping for more AI Brief stories.