TrendAI is enhancing its collaboration with Nvidia to fortify security measures for agentic AI. This effort centers around Nvidia OpenShell, an open-source runtime specifically designed for autonomous AI agents.
This integration ensures that security is embedded within the runtime architecture that executes agents capable of planning, utilizing tools, and operating over extended periods. Unlike traditional systems, agentic AI often operates across varied environments and can take actions independently of direct user prompts.
TrendAI, the enterprise division of Trend Micro, is set to implement a robust security framework known as TrendAI Vision One for OpenShell. This architecture will feature governance controls and monitoring systems that enable agents to adapt their behavior over time.
Shift In Risk
Enterprises are increasingly interested in agentic AI for automating workflows across diverse applications, data sources, and infrastructures. However, this advance introduces a new set of security concerns. Traditional AI security has primarily concentrated on short, interactive sessions between a user and a model. In contrast, agentic AI systems function more like ongoing software processes capable of invoking tools and taking actions automatically.
Nvidia promotes OpenShell as a runtime for “long-lived, self-evolving agents” that possess capabilities in planning, memory storage, and tool execution. This design significantly broadens the spectrum of potential risk scenarios, including unauthorized skills, concealed behaviors, prompt injections, and unintentional access to systems.
Rachel Jin, Chief Platform and Business Officer and Head of TrendAI, emphasized that systems capable of autonomous action present a unique security landscape.
“Agentic AI alters the security landscape. When AI systems can plan, act, and independently interact with other tools, the risk profile diverges greatly from that of traditional AI. Our partnership with NVIDIA enables us to integrate security directly into the architecture, granting organizations the visibility and control they require for adopting agentic AI,” Jin remarked.
Runtime Controls
TrendAI’s strategy encompasses three critical phases: before, during, and after the agent execution. This includes establishing trust boundaries, enforcing policies during runtime, and ensuring continuous monitoring of agent activities.
The partnership is framed as part of a larger initiative to make the deployment of agentic AI more manageable in large organizations, where governance, audit, and security processes are essential once systems progress beyond the pilot phase.
Pat Lee, Vice President of Strategic Enterprise Partnerships at Nvidia, stated that the collaboration seeks to enhance visibility and control for developers working with autonomous agents.
“Agentic AI opens the door to a new class of applications capable of planning, reasoning, and taking action. By collaborating with TrendAI, we are enabling developers to incorporate visibility and control, thereby making the deployment of autonomous AI safer,” Lee noted.
Broader Integration
This expanded collaboration goes beyond OpenShell to encompass additional Nvidia agentic AI initiatives, such as the Nvidia AI-Q blueprint and the Nvidia NeMo Agent Toolkit. TrendAI asserts that this alignment cultivates a uniform approach to security, governance, and observability as agentic systems proliferate across enterprise environments.
The security architecture embedded in TrendAI Vision One for OpenShell includes centralized governance and compliance controls within the agent runtime. Additionally, it scans agent skills and Model Context Protocol integrations, which are frequently utilized to link agents with external tools and data sources.
The system also conducts behavioral analysis to detect hidden or malicious activities. Inline policy enforcement is implemented to block any untrusted skills and actions during runtime.
TrendAI has also highlighted AI-specific threat protections, including detection capabilities for prompt injections and sensitive data exposure. Monitoring and audit functionalities encompass agent telemetry and SIEM integration, which security teams utilize for centralized logging, alerting, and incident response.
Enterprise Adoption
Organizations embracing agentic AI frequently require controls that align with existing risk management protocols. Security teams typically prioritize policy enforcement, activity monitoring, and documentation for audits and investigations. Developers also seek frameworks that minimize friction when agents are updated, new tools are connected, or workloads transition between different environments.
TrendAI positions its security layer to govern agent access to tools and to detect and enforce risk management throughout an agent’s operational lifecycle. This initiative forms part of a broader security model that encompasses AI infrastructure, data pipelines, models, and applications.
Drawing upon decades of cybersecurity expertise and threat intelligence, TrendAI is supported by global security operations and research that track vast volumes of threats daily.
As organizations evaluate OpenShell and associated Nvidia tools, they are likely to focus on how seamlessly security controls integrate into developer workflows, how policies can be efficiently managed at scale, and what telemetry is accessible to security operations teams as agentic AI advances from experimentation to full-scale deployment.
