多智能体系统风险与控制：跨组织边界部署AI智能体的分析框架
原标题：Risks and Controls for Multi-Agent Systems: an analytical framework for deployment of AI agents across organisational boundaries
AI 摘要
澳大利亚工业、科学和资源部委托Gradient Institute撰写了一份关于跨组织边界部署AI智能体的多智能体系统风险与控制分析框架报告。报告由Alistair Reid等人编写，旨在为AI智能体的部署提供风险评估和控制建议。该报告是此前关于受治理的基于LLM的多智能体系统风险分析技术的后续报告。
正文节选
[Page 1] Risks and controls for multi-agent systems An analytical framework for deployment of AI agents across organisational boundaries 10 August 2026 Prepared by Alistair Reid, Simon O'Callaghan, Dustin Venini, Liam Carroll and Tiberio Caetano of Gradient Institute for the Department of Industry, Science and Resources industry.gov.au Risks and controls for multi-agent systems | industry.gov.au 1 [Page 2] Acknowledge