Open Data

ServiceNow and Nvidia say their new open-source model is built for security - here's why

ServiceNow and Nvidia say their new open-source model is built for security - here's why

ServiceNow and Nvidia have collaborated to release Apriel 2.0, an open-source AI model designed to help enterprises build custom, secure AI agents. This development addresses the growing demand for specialized tools that can handle multistep tasks with minimal human oversight, moving beyond simple chatbot interactions. By leveraging Nvidia’s Nemotron architecture and integrating seamlessly with ServiceNow’s platform, the model allows developers to access underlying code, fostering innovation while ensuring that organizations can tailor AI solutions to their specific workflows and data requirements. A primary advantage of this open-source approach is the enhanced security and transparency it offers compared to proprietary, general-purpose alternatives. The model operates within compliance-certified infrastructure, automatically inheriting existing organizational permissions and audit trails. This feature is particularly critical for federal agencies and regulated industries, such as healthcare and finance, which require strict data governance. By providing safety guardrails and responsible data vetting, the technology aims to mitigate the privacy and security risks often associated with handing over proprietary data to external AI systems. This launch is highly relevant to open_data as it demonstrates how open-source frameworks can solve real-world enterprise challenges like trust and compliance. While many businesses struggle to see a return on investment from AI initiatives, offering transparent, customizable models empowers organizations to control their data and build trust. As the market shifts toward automated decision-making processes, open-source solutions provide a pathway for responsible adoption, ensuring that the benefits of advanced AI do not come at the cost of security or regulatory violation.

Source: zdnet.com
Published on 2025-10-29

Related news