DigestAI news desk

Cut through the AI noise.

Listen
Enterprise & Industryupdated 3 min read

Anthropic and NVIDIA launch Open Agent Safety Platform for AI control

NVIDIA announced the Open Agent Safety Platform, an open software system designed to enhance AI security. Anthropic collaborated on this initiative to add security layers to its agent stack. The platform combines Claude Managed Agents, a suite of APIs for deploying production-grade agents, with NVIDIA's open-source OpenShell software. This approach addresses the growing risk as companies deploy…

32 sources primary source HN 142

Key points

  • NVIDIA announced the Open Agent Safety Platform with Anthropic collaboration.
  • Claude Managed Agents holds credentials in a vault so agents never see them.
  • NVIDIA OpenShell is open source under Apache 2.0 license on GitHub.

The system uses a layered defense strategy. Claude Managed Agents runs the agent loop on a separate server from the sandbox, keeping credentials in a vault so the agent never sees them. It also provides audit trails and integrates with existing access controls. NVIDIA OpenShell acts as a secure runtime that governs agent behavior. It blocks all actions unless explicitly allowed by rules, checking tools, files, and network connections. OpenShell logs every decision and uses a policy prover to mathematically confirm the agent's reach under defined rules.

Claude Managed Agents is available today, supporting long-running sessions and multi-agent orchestration. NVIDIA OpenShell is open source under the Apache 2.0 license. Companies like Notion, Rakuten, and Asana are using these tools to deploy agents for tasks ranging from code shipping to project management. The collaboration aims to give companies more control over what their AI agents can execute and reach.

The story so far

8 episodes →
  4. Anthropic and NVIDIA launch Open Agent Safety Platform for AI controlthis story
Full story from Anthropic Engineering primary sourceOpen source ↗

Giving companies more control over their AI agents, with NVIDIA

Anthropic Engineering · 28 September 2026

Loading the full article…

This text was published by Anthropic Engineering. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Coverage and discussion

32sources
Topics · follow one to build your own front page
NVIDIAAnthropicNotionRakutenAsanaClaude

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Enterprise & Industry

All →

Related stories

hot Introducing Mistral Large 4 new Cyara’s CEO outlines AI agent oversight as a workforce model hot ChatGPT generates New Yorker-style cartoons with real cartoonists' signatures without permission new Arcana adds institutional portfolio data to Claude’s connectors hot OpenAI's GPT-6 Astra: Changing Work and Research new Opinion: open-weight AI models may worsen cyber risks without banning closed APIs hot HubSpot adds AEO tool to track brand mentions in AI answers new OpenAI reportedly in talks for $30B round with UAE funds