DeepMind page signals funding call for multi-agent AI safety

A Google DeepMind page says Google DeepMind and partners announced a multi-agent safety research funding call, while Schmidt Sciences program text frames the problem as safety for coordinated networks of specialized AI systems rather than only single models. The Schmidt text also ties the call to Google DeepMind’s “Distributional AGI Safety,” ARIA’s “Scaling Trust” thesis, and the Cooperative AI Foundation’s “Multi-Agent Risks from Advanced AI” report, making the program a source-backed marker for growing institutional attention to agent coordination. The supplied readable evidence does not verify the $10 million figure, the exact deadline, or each institution’s role, so those details should remain open until fuller official materials are checked.

AI · June 12, 2026

Sources & context

Evidence, open claims, and reader-facing editor notes.

Dek

A $10 million fund for multi-agent AI safety would turn agent interaction, collusion, conflict, and oversight into an institutional research lane as AI products move from single assistants toward networks of autonomous actors.

Synthesis angle

Investigate how funders are trying to define safety infrastructure for multi-agent AI systems before deployment standards and liability rules harden.

Source

$10M Fund Targets Safety of Multi-Agent AI Systems

Preferred source plan

Google DeepMind announcement
Schmidt Sciences program page
Cooperative AI Foundation grant materials
ARIA program materials
Google.org funding statement
Grant application criteria and funded research agenda

Reading the dossier...

DeepMind page signals funding call for multi-agent AI safety | Italo Daily