TrendAI Extends NVIDIA Agent Safety Platform with Threat Intelligence and AI Factory Security

TrendAI and NVIDIA Partner to Secure Autonomous AI Agents

TrendAI announced its support for the NVIDIA Agent Safety Platform1

1

, marking a significant collaboration to address security challenges as enterprises deploy autonomous AI agents at scale. The partnership integrates TrendAI Vision One with NVIDIA's full-stack infrastructure, combining hardware-level detection capabilities with comprehensive threat intelligence2

2

. This integration addresses a critical enterprise need: securing AI agents that hold credentials, maintain context across sessions, and act on production systems without human approval at each step.

The collaboration tackles what both companies identify as a fundamental security principle for agentic AI—no single control makes an agent secure1

1

. Security must be embedded into every layer the agent depends on, from the model that reasons to the harness that orchestrates sub-agents and the tools and data the agent can access. As organizations transition from pilot programs to production-scale agent fleets spanning multiple teams and business units, security has become the deciding factor in whether agentic AI reaches full deployment.

Multi-Layered Approach to AI Factory Security

The NVIDIA Agent Safety Platform establishes authority in the infrastructure beneath the agent through OpenShell, which enforces policy outside the agent's execution environment where it cannot be bypassed through prompt manipulation1

1

. NVIDIA BlueField-4 DPUs and NVIDIA DOCA add a host-independent security layer implemented in silicon, creating a hardware-rooted foundation that remains trustworthy even if the host operating system is compromised.

TrendAI Vision One complements this enforcement infrastructure by providing the intelligence layer that informs policy decisions, visibility into agent activities, and threat detection and response capabilities1

1

. "Agentic AI is the most significant shift in enterprise technology in a generation, and security needs and guardrails are in place to keep the agents aligned with the intentions of the organization," said Sharda Tickoo, Country Manager for India and SAARC at TrendAI1

1

. The solution extends safeguards across the entire AI factory, covering agents, models, containers, storage, and networks on a unified platform.

Comprehensive Protection Across Agent Components

TrendAI Vision One delivers model protection through inline inspection of prompts and responses, actively blocking prompt injection attacks and jailbreaks while redacting sensitive data before it leaks1

1

. This addresses one of the most pressing vulnerabilities in AI systems where malicious actors attempt to manipulate agent behavior through carefully crafted inputs.

For harness and tool governance, security teams gain control over which tools and MCP servers agents can access, with the ability to inspect every tool call and analyze how skills and AI artifacts behave at runtime2

2

. This runtime behavior analysis proves crucial as agents increasingly operate with expanded permissions and access to production systems.

Data and Identity Controls for Enterprise Scale AI

The platform implements data and identity controls that find and classify sensitive data feeding AI systems, preventing agents from accessing unauthorized data sources1

1

. TrendAI Vision One extends file security and data loss prevention to files that agents read and write, while stripping excess privilege from agents and other non-human identities. This addresses the challenge of privilege creep as agent fleets expand across organizations.

Using NVIDIA BlueField DPU and DOCA telemetry, TrendAI Vision One maintains detection capabilities that remain trustworthy even when the host OS is compromised2

2

. The system correlates AI signals with endpoint, network, and identity telemetry to ensure a hijacked agent cannot move laterally through enterprise systems undetected. This hardware-level detection paired with Zero Day Initiative threat intelligence creates a security foundation designed to scale with enterprise AI deployments while maintaining the visibility and control that security teams require for production environments.