【文章标题】：Agentic Trust Controls
代理信任控制
【文章正文】：
Loading…
加载中…
Extend the security, privacy, and AI governance frameworks you already use with controls designed specifically for agentic AI.
通过专为代理型AI设计的控制措施，扩展您现有的安全、隐私和AI治理框架。
Frameworks like ISO 27001 and ISO 42001 provide strong foundations for governance, but they were not intended to cover every emerging technology.
诸如ISO 27001和ISO 42001等框架为治理提供了坚实基础，但它们并非旨在覆盖所有新兴技术。
They expect organizations to define additional controls based on their own risks. The challenge? Most teams don’t have a practical starting point for governing AI agents.
这些框架期望组织根据自身风险定义额外控制措施。但挑战在于？大多数团队缺乏治理AI代理的实际起点。
Agentic Trust Controls fills that gap with an open source set of controls designed for the unique risks of agentic AI. No need to start from scratch. Simply extend the frameworks you already use with controls that are ready to turn on.
代理信任控制通过一套开源控制措施填补了这一空白，专为代理型AI的独特风险设计。无需从零开始，只需用现成的控制措施扩展您已有的框架。
Get a baseline of 43 controls covering identity and authority, action guardrails, memory protection, instruction integrity, adversarial testing, and runtime instrumentation.
获取涵盖身份与权限、行为护栏、内存保护、指令完整性、对抗测试和运行时检测的43项基线控制措施。
Get 22 deployment-focused controls covering agent intake and inventory, credential control, oversight staffing, vendor review, runtime monitoring, and staff training.
获取22项以部署为重点的控制措施，涵盖代理引入与清单、凭证控制、监督人员配置、供应商审查、运行时监控和员工培训。
Are you building agents or deploying them?
您正在构建代理还是部署它们？
Choose the framework you already work in, and then map the controls.
选择您已在使用的框架，然后映射控制措施。
The exported controls are mapped and ready to be put in scope.
导出的控制措施已映射完毕，随时可纳入范围。
Access Agentic Trust Controls through our hosted MCP server so AI agents and tools can reference the controls directly. (Coming soon.)
通过我们托管的MCP服务器访问代理信任控制，使AI代理和工具能直接引用这些控制措施。（即将推出）