DeepMind and Partners Announce Up to $10M Funding Call for Multi-Agent AI Safety Research
weballergy · x · 2026-08-21
Google DeepMind, together with Schmidt Sciences, the Cooperative AI Foundation, and ARIA, supported by Google.org, has announced a technical research funding call of up to $10M for researchers worldwide on multi-agent AI safety.
Context and motivation:
- A decade of safety work focused on individual models, but AI is entering an era where millions of agents built by different organizations will communicate, negotiate, and transact across digital environments.
- Large interacting agent groups can suddenly produce emergent collective behaviors that current tools cannot predict, measure, or monitor; most safety evaluations still analyze models in isolation.
- The call targets how large-scale multi-agent systems behave as groups and frameworks to understand and mitigate risks — the "invisible" safety risks arising when independent systems interact.
The poster (@weballergy) notes this is an active research area on their end, invites the broader community to work on it, and admits no real-world system fully meets the spec yet — many subproblems remain, especially in valuable real-world scenarios.
More from Safety
- ALIFE2026 Workshop Focuses on Agentic AI Safety — danbri · 2026-08-21
- OpenAI monitoring adds 20% inference overhead, reshaping agent economics — Crescitaly · 2026-08-21
- Auditing safety signals in Zero Data Retention without provider access — Crescitaly · 2026-08-21
- Stop parenting your AI: treating it as a collaborator improves output — incajb · 2026-08-21
- Critical Microsoft Copilot CoSnitch Vulnerability Exposes Data via Single Click — emmanuelvivier · 2026-08-21
- Multi-agent safety research lacks fully realized real-world examples — weballergy · 2026-08-21