Overview
- Palo Alto Networks, which announced the releases Monday, introduced Prisma AIRS 3.0 to secure AI agents across design, deployment and runtime.
- The platform adds agent discovery across cloud, SaaS and endpoints, scans agent components for flaws through Agent Artifact Security, and runs AI red teaming to simulate attacks and set runtime policies.
- An AI Agent Gateway enters limited preview as a central control point for agent identity, runtime enforcement and oversight, while deeper endpoint protections are planned after a proposed Koi acquisition closes.
- A major update to Prisma Browser turns the web into a guarded workspace for AI agents by auto-detecting AI use, blocking data from flowing to unmanaged tools, and stopping prompt injection, where hidden website text tries to hijack an agent.
- Prisma SASE extends these controls with identity-based policies and data protection across networks and apps, as partners like Accenture and ServiceNow say the tools give enterprises the guardrails needed to scale agent-driven work.