Nvidia outlines AI agent security framework with OpenShell

Nvidia has outlined a security framework for AI agents and highlighted OpenShell as a runtime for enforcing controls. Its approach centres on enforceable boundaries, accountable owners and evidence that protections work.

The framework sets out how organizations should secure AI systems that can reason, use tools and adapt their actions based on the data they encounter. It argues that the same core responsibilities used in internet and cloud security still apply, including identity management, access control, limiting exposure and verifying that safeguards are effective.

Rather than treating AI security as a separate discipline, Nvidia presents it as an engineering task that requires defined requirements, clear ownership and testing.

A central point in the framework is that security depends on the full agent stack, not just the model. Nvidia describes that stack as including models, harnesses that organize context and workflows, and the runtime environments where actions are executed, with each layer carrying its own security responsibilities.

One example in the framework involves an AI agent updating a customer record after encountering malicious instructions in an attached document. In that scenario, a network policy should block any attempt to export customer data to an unauthorized destination, while protected logs should record the attempted tool call, the authorization decision and the outcome.

Nvidia also argues that permission to update a customer record should not automatically allow an agent to export the same data elsewhere. If an agent needs extra access, it can request it, but it should not be able to grant that access to itself.

Security boundaries must remain effective even when an agent makes the wrong decision, which makes the runtime environment critical, since it should impose limits on files, network destinations and processes independently of the agent's reasoning. OpenShell is presented as Nvidia's open-source runtime for that purpose, providing sandboxed execution and enforcing policy controls outside the agent's reach while governing access to data, networks and system resources.

In the wider ecosystem, Open Secure AI Alliance partners are building on OpenShell, with Cisco's DefenceClaw adding a governance layer and JFrog integrating scanning and verification of agent skills.

As businesses pursue AI-driven productivity gains while governance and security practices are still taking shape, the framework's aim is to secure agents from testing through deployment. What to watch next: OpenShell's adoption, the maturity of alliance integrations, and whether such runtime controls become an industry norm.