提出安全框架，防止AI客服组合后出现意外危险行为
From Workflow Automation to Capability Closure: A Formal Framework for Safe and Revenue-Aware Customer Service AI
- 构建形式化框架，检测多AI代理组合时的潜在安全风险
- 发现两个独立安全的代理组合后可能触发禁止目标
- 适合关注AI客服系统安全与商业合规的开发者和管理者
客户服务中心自动化正经历结构性变革，主导范式从脚本化聊天机器人和单智能体响应转向由专业化AI代理组成的网络，这些代理可动态组合账单、服务提供、支付和履约等能力。这一转变引入了当前平台尚未解决的安全缺口：两个单独验证为安全的代理，在组合后可能通过彼此独立不存在的协同依赖关系，达成禁止目标。
原文摘要 · Abstract (English)
Customer service automation is undergoing a structural transformation. The dominant paradigm is shifting from scripted chatbots and single-agent responders toward networks of specialised AI agents that compose capabilities dynamically across billing, service provision, payments, and fulfilment. This shift introduces a safety gap that no current platform has closed: two agents individually verified as safe can, when combined, reach a forbidden goal through an emergent conjunctive dependency that neither possesses alone.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。