From Nvidia: 2025-04-28 11:00:00

Agentic AI is transforming cybersecurity, offering new ways to secure AI systems and defend against new risks. This shift requires a fundamental change in cybersecurity strategies to leverage AI agents for threat detection, response, and security operations.

AI agents can autonomously investigate security vulnerabilities, prioritize findings, and optimize security alert triaging. Leading organizations like Deloitte and AWS are using NVIDIA AI Blueprint to accelerate software patching and vulnerability management, enhancing security operations center efficiency.

To protect agentic AI applications, organizations must conduct pre-deployment testing and establish runtime controls. Tools like Garak and NVIDIA NeMo Guardrails help identify vulnerabilities, enforce policy boundaries, and ensure consistent and safe agent behavior during execution.

NVIDIA Confidential Computing provides runtime protections for sensitive data and agent actions, reducing the risk of exposure during training and inference. Major service providers like Google Cloud and Microsoft Azure offer this technology to safeguard data in use across various deployment environments.

The NVIDIA AI Enterprise software platform ensures API stability, vulnerability management, and secure deployment of AI components. Container signatures, model signing, and software bill of materials enable verification of AI components in the supply chain, enhancing trust and security in agentic AI systems.

Agentic AI infrastructure must enforce isolation, visibility, and control to secure autonomous AI systems effectively. NVIDIA BlueField DPUs and DOCA Argus framework provide real-time visibility, memory forensics, and security controls to address potential threats and reinforce a comprehensive security architecture.

Integrators use NVIDIA Confidential Computing to enhance security in agentic infrastructure, with technologies like Protected PCIe enabling scalable workload deployment while maintaining security. These components support local and remote attestation for verifying platform integrity before deploying sensitive workloads.

Secure AI infrastructure is crucial for AI factories and cyber-physical systems, where agentic AI powers automation and real-world decision-making. Partners like Cisco are integrating NVIDIA BlueField DPUs to deliver secure and efficient AI deployments for enterprises, safeguarding critical infrastructure against cyber threats.



Read more at Nvidia: How Agentic AI Enables the Next Leap in Cybersecurity