重新思考代理安全：作为网络问题

AI 摘要

本文提出将AI代理安全视为网络问题，主张采用确定性执行与上下文感知过滤相结合的网络架构，以解决现有代理中心防御的局限性。作者认为，依赖LLM推理的防御易受提示注入等攻击，无法提供强安全保证，而网络侧控制可提供更可靠的保障。文章还提出了参考架构，并指出了未来研究方向。

正文节选

Rethinking Agent Security as a Networking Problem Abstract. AI agents are rapidly becoming more capable and widely deployed, promising substantial gains in productivity and enabling new classes of applications. However, their growing autonomy also introduces significant privacy and security risks. Existing defenses are predominantly agent-centric, relying on the agent itself to detect threats and enforce privacy and security policies. This approach is fundamentally limited because it entrusts po

发布时间：2026-08-13 12:00

抓取时间：2026-08-13 12:24

来源机构：arXiv