IndustryMulti-source

NVIDIA AI Podcast 复播:Thom Wolf 谈开源模型与 Agent 安全

Why we picked it

NVIDIA 播客复播第一期就请来了 Hugging Face 的 Thom Wolf,聊 Agent 怎么配独立身份和权限,还提到 NVIDIA Sentry。做 Agent 开发、关心安全落地的可以听听。

NVIDIA 宣布 AI Podcast 复播,首期嘉宾是 Hugging Face 联合创始人 Thom Wolf 与 ServiceNow 的 Amanda Grady。节目共 14 个章节,时长约 15 分钟,话题覆盖开源与专有模型在网络防御中的应用、给 Agent 配置独立身份与权限机制等。对话还介绍了 AI Passport 与 NVIDIA Sentry 两个安全相关项目,讨论了安全能力提升后能为开发者解锁哪些场景。

Body

播客还延伸到开源与安全的关系、如何把安全机制带入物理世界,以及降低微调门槛的具体做法。Thom Wolf 作为 Hugging Face 的开源生态代表,视角侧重开放模型在安全体系中的角色。整期内容面向想在 Agent 场景中落地权限与安全控制的开发者。

Original · NVIDIA

The AI Podcast is back.

@huggingface's @Thom_Wolf and @ServiceNow's Amanda Grady join us to talk open models, agent safety, and helping more people build with AI.

0:00 AI safety beyond the prompt 1:50 Open and proprietary models for cyber defense 3:28 Giving agents their own identity and permissions 5:07 AI Passport and NVIDIA Sentry 6:03 What better security unlocks 8:32 Open source and security 10:58 Bringing safety into the physical world 13:10 Making fine-tuning more accessible 14:09 What’s next—and why they’re optimistic

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include NVIDIA, Hugging Face, agent security, open-source models.

Source attribution: curated from NVIDIA (@nvidia). Cross-source corroboration: 11 independent sources have covered this development (including techcrunch, Tech in Asia, IT之家), confirming industry-wide signal strength.