Nvidia has revealed three fresh microservices under its NIM (Nvidia Isaac and Jarvis Modules) umbrella, aimed at catalyzing businesses to adopt AI agents by bolstering security and control measures.
The first among these is a content safety feature designed to circumvent potentially harmful or biased AI-driven outputs. The second microservice is to ensure that AI agent-oriented discourse stays within approved topics. The final service is aimed at defending against jailbreak attempts, i.e., attempts to break software restrictions.
These new add-ons to the Nvidia NeMo Guardrails, an established open-source arsenal of tools targeted at enhancing corporate AI applications, are expected to address potential “blind spots” in AI security by using multiple lightweight, specialized models.
The release of these microservices comes amid signs of slower-than-expected enterprise uptake of AI agent technology. Experts have forecasted one billion AI agents running within a year, but actual rates of adoption may prove more modest.
According to a recent Deloitte report, only 25% of enterprises either utilize AI agents presently or plan to by 2025, with 50% expected to have embraced this technology by 2027. Nvidia’s new initiatives may eventually succeed in convincing enterprises of the security and dependability of AI agents. Only time will reveal the actual outcome.
Original source: Read the full article on TechCrunch
