|
Cloudflare TV

Securing agentic harnesses: Control, visibility, and safe tool use

主講人:Gérard Zapf
最初直播時間:10月24日,下午3:00 - 下午4:00 [EDT]
As organizations build agentic systems that call tools, retrieve data, and take action, security must move beyond prompt-level controls. This session explores how to secure agentic harnesses with the visibility, policy, and access controls needed to make tool-using AI safe and governable in production. Learn to implement authentication, authorization, logging, and policy enforcement across model calls, tools, and retrieval flows — keeping agents useful without giving them unchecked reach. Also learn practical patterns for controlling agent behavior, reducing risk from unsafe tool use, and making agentic systems auditable and operationally defensible.
English

文字記錄 (測試版)

background