Nvidia and Cisco Introduce New Tools to Enhance AI Safety and Security

Nvidia debuts NeMo Guardrails microservices for AI agent control, while Cisco unveils AI Defense tools to address enterprise security challenges.

Nvidia has launched three NeMo Guardrails Inference Microservices (NIMs) focused on content safety, topic control, and jailbreak detection to enhance AI agent reliability and security.
The Content Safety NIM aims to block harmful or biased outputs, while the Topic Control NIM ensures conversations stay on approved topics, and the Jailbreak Detection NIM identifies attempts to bypass AI restrictions.
These tools are optimized for Nvidia GPUs and designed to minimize latency while providing robust protection, with open-source options available for developers.
Cisco announced its AI Defense suite, including tools to validate AI model performance, detect unauthorized AI applications, and deploy guardrails to prevent misuse of AI systems.
Both companies aim to address growing enterprise concerns over AI safety, data privacy, and governance as agentic AI adoption continues to expand.