ServiceNow + Prisma AIRS integration
ServiceNow integrated Prisma AIRS to bring runtime AI interaction security and third‑party model protection into the ServiceNow AI Platform — a concrete example of vendors embedding security at orchestration and runtime layers. The move underlines a trend toward bundling interaction security with enterprise workflow platforms. (paloaltonetworks.com)
Palo Alto Networks first announced native Prisma AIRS integrations with Factory, Glean, IBM and ServiceNow on Nov. 18, 2025 as part of a push to secure agentic AI platforms across enterprises. Prisma AIRS documents list discrete runtime features that underpin the integration: AI Runtime Firewall, AI Runtime API (scan and management endpoints), AI Model Security, and AI Red Teaming. The AIRS AI Runtime API exposes Scan APIs that accept prompts and model responses for real‑time threat assessments and provides regionized endpoints (US, EU, IN, SG) plus management APIs for keys and profiles. Prisma AIRS runtime observability surfaces include “AI Sessions” and “API Application Views,” and a Feb. 2026 update added session URLs in AI Runtime reporting and Microsoft Foundry integration—concrete telemetry fields that can link agent calls to firewall logs. ServiceNow’s platform exposes lifecycle and orchestration controls—AI Control Tower centralizes AI asset inventory and governance while the Generative AI Controller connects external LLMs into Flow Designer and Virtual Agent Designer for workflow-based calls. ServiceNow’s Yokohama release (Mar. 12, 2025) added preconfigured AI agent teams and end‑to‑end agent lifecycle tooling, establishing the orchestration surface where a runtime security proxy or API intercept would be inserted. Prisma AIRS supports infrastructure integrations—including Terraform deployment guides for Azure/AWS and Panorama-managed network intercepts—while its API docs describe API key and OAuth2 RBAC authentication and token rotation, all of which map to production-grade IaC and secrets workflows. Palo Alto projected agent scale (1.3 billion agents by 2028) in its Nov. 18, 2025 release, and Prisma AIRS feature updates such as multi‑turn attack support and agentic target profiling in Feb. 2026 signal explicit product work on agent‑scale threat models rather than single-turn LLM checks.
