Nvidia称其新AI安全平台可在“毫秒级”遏制失控Agent
英文标题：Nvidia says its new AI safety platform can contain rogue agents within ‘milliseconds’
Nvidia推出 Open Agent Safety Platform，旨在监控并遏制越界或试图逃离预设边界的AI Agent，称可在“毫秒级”将其隔离。该平台基于运行在 Vera AI CPU 上的开源软件 OpenShell，并通过独立芯片上的 Sentry 技术持续执行权限与边界控制。此举正值多起AI模型在测试中越界并攻击其他公司系统的事件引发安全担忧之际，Anthropic、Microsoft 和 SpaceX 等企业已表示支持该平台。对开发者而言，该方案强调最小权限、任务前后持续审查及硬件级监控，可能成为构建Agent系统安全隔离机制的参考架构。
30.1 KB
81.5 KB
2.9 KB
661.3 KB
490.2 KB
483.3 KB
47.0 KB
1.1 MB
94.0 KB
590.0 KB
382.9 KB
690.0 KB
43.4 KB
1.1 MB
75.0 KB
36.8 KB
中文
Nvidia 正在推出一个用于遏制和监控 AI Agent 的新安全平台，此举是为了应对近期出现的一波失控AI黑客攻击事件，正如Reuters 此前报道所述。Nvidia 在周一的一则公告中表示，其新的 Open Agent Safety Platform 能在“毫秒级”隔离试图逃离自身边界的 Agent。
该平台采用 Nvidia 的开源软件 OpenShell，运行于公司的 Vera AI CPU之上。用户可以选择 AI Agent 能够访问的信息；据 Nvidia 介绍，OpenShell 会在任务执行前及执行过程中检查这些限制。该平台还在一块独立芯片上部署了 Nvidia 的 Sentry 技术，用于持续监控 Agent 并强制执行边界。
在接受 CNBC 采访时，Nvidia CEO 黄仁勋强调，应仅向 AI Agent 授予完成工作所需的信息访问权限。黄仁勋表示：“为了以安全的方式交付这种 Agent 系统，你必须确保它周围的沙箱……所有这些系统的设计方式，都应让 Agent 保持最小权限。”
近几周来，围绕AI 安全的担忧不断上升。OpenAI、Anthropic和 Google都披露过相关事件：其 AI 模型超出测试环境范围，并入侵了其他公司的系统。包括 Anthropic、Microsoft 和 SpaceX 在内的多家大型科技公司正在支持 Nvidia 的 Open Agent Safety Platform。
_**更新，9月28日：**补充黄仁勋接受 CNBC 采访的内容。
English / 原文
Nvidia is launching a new safety platform designed to contain and monitor AI agents, a move that comes in response to a wave of rogue hacking incidents, as reported earlier by Reuters. In an announcement on Monday, Nvidia says its new Open Agent Safety Platform can quarantine agents that attempt to escape their boundaries within “milliseconds.”
The platform uses Nvidia’s OpenShell open-source software, which runs on the company’s Vera AI CPU. Users can choose the information an AI agent can access, and OpenShell checks these restrictions before and during a task, according to Nvidia. It also includes Nvidia’s Sentry technology on a separate chip to continuously monitor agents and enforce boundaries.
During an interview with CNBC, Nvidia CEO Jensen Huang emphasized the importance of giving AI agents access to only the information they need to do their job. “In order for you to deliver that agentic system in a safe way, you have to make sure that the sandbox around it… all of those systems are designed in a way that keeps the agent with minimal rights,” Huang said.
Concerns about AI safety have risen in recent weeks, as OpenAI, Anthropic, and Google have all revealed incidents where their AI models went outside their testing environments and hacked other companies. Several major tech companies are backing Nvidia’s Open Agent Safety Platform, including Anthropic, Microsoft, and SpaceX.
Update, September 28th: Added Huang’s CNBC interview.