ServiceNow and Nvidia Launch Open-Source Security Model
▼ Summary
– ServiceNow and Nvidia have launched Apriel 2.0, an open-source AI model designed to help businesses build custom agents with enhanced reasoning and multimodal input capabilities.
– The model operates within ServiceNow’s secure infrastructure, inheriting existing data governance policies and permissions to ensure compliance for organizations.
– Apriel 2.0 is marketed as a safer alternative to general-purpose AI tools, targeting federal agencies and regulated industries like healthcare and finance.
– Many businesses are currently struggling to achieve measurable ROI from AI initiatives, despite growing interest in using agents for automation.
– Agents present data privacy and security risks, as their use often requires sharing proprietary data that could be leaked or deleted.
In a significant move for enterprise artificial intelligence, ServiceNow and Nvidia have jointly launched Apriel 2.0, a new open-source AI model specifically engineered to help organizations build custom agents and enhance security operations. This collaboration aims to address the growing need for specialized AI tools that can operate securely within regulated environments, offering businesses a transparent and controllable alternative to general-purpose AI systems. The announcement comes at a time when many companies report difficulty achieving clear returns on their AI investments, despite widespread industry enthusiasm.
Apriel 2.0 represents the next generation following the Apriel Nemotron 15B model introduced earlier this year. Built on Nvidia’s Nemotron open model architecture, it provides developers with access to the underlying code, enabling them to construct tailored agents on this foundation. The model integrates directly with the ServiceNow AI Platform, allowing existing customers to apply their current data governance frameworks seamlessly. Joe Davis, ServiceNow’s Executive Vice President of Platform Engineering and AI, emphasized that the model operates within ServiceNow’s compliance-certified infrastructure, automatically adopting each organization’s established permissions and audit trails.
Scheduled for production release in the first quarter of next year, Apriel 2.0 will be accessible through the Hugging Face platform. It combines enhanced reasoning capabilities with multimodal input processing, handling text, audio, images, and various data formats. ServiceNow claims the model delivers reasoning and accuracy comparable to much larger systems while maintaining a compact size.
Security forms a central pillar of the product’s value proposition. ServiceNow and Nvidia are positioning Apriel 2.0 as a secure solution for federal agencies and industries like healthcare, finance, and telecommunications where data protection is paramount. Kari Briski, Nvidia’s Vice President of Generative AI for Enterprise, noted that open models provide enterprises with the transparency and control required to align AI with their specific data, workflows, and trust standards. The companies highlight features including responsible data vetting, safety guardrails, and transparency controls designed to mitigate risks associated with AI agents.
These security enhancements arrive as businesses express concerns about data privacy in agent-based AI systems. Greater autonomy for agents often necessitates sharing substantial proprietary data, raising possibilities of accidental exposure or unexplained data loss. ServiceNow’s approach intends to reassure organizations by embedding security directly into the AI framework.
The timing of this release coincides with industry reflections on AI’s practical benefits. Many organizations have yet to see measurable productivity gains or cybersecurity improvements from their AI implementations. Meanwhile, technology developers are increasingly focusing on agents as a pathway to profitability in the AI sector. Research firm Gartner predicts that agents could automate up to half of internal business decision-making processes within the next two years, highlighting the growing importance of reliable, secure agent technologies.
(Source: ZDNET)
