
Palo Alto Networks has fired the starting shot in the race to secure the global AI boom with the launch of Prisma AIRS 2.0 — a sweeping platform upgrade designed to protect every layer of enterprise AI, from the autonomous agents that execute tasks to the models powering them.
As companies sprint toward AI transformation, a staggering gap remains: while 78 percent of enterprises are already using AI, only 6 percent have the guardrails to do it safely. Prisma AIRS 2.0 steps in as the industry’s most comprehensive answer to that problem.
Unveiled in Singapore, the enhanced platform completes the integration of Protect AI, bringing together deep model inspection, real-time agent defense, and nonstop autonomous red-teaming into a single, unified system. For organizations worried about the spiraling risks of unchecked AI adoption, Palo Alto Networks positions AIRS 2.0 as the fuel that turns security from a bottleneck into a catalyst for innovation.
Anand Oswal, Executive Vice President for Network Security, put it simply: AI is reshaping enterprises at lightning speed, and the security world must keep pace. With AIRS 2.0, he said, Palo Alto Networks is redefining what it means to secure AI at scale—protecting models, agents, data, and users in every interaction.
The platform closes critical gaps across the AI lifecycle by connecting model and agent inspection during development with real-time defenses at runtime, all validated through continuous autonomous red-teaming. Already adopted by major players in finance, healthcare, and government, Prisma AIRS 2.0 now expands its capabilities with three upgraded protection layers built for modern AI threats.
AI Agent Security cracks down on the explosion of autonomous systems—whether approved or rogue—by detecting prompt injections, tool misuse, and malicious agent behavior as they happen. AIRS also maps every AI agent in an enterprise, finally giving security leaders control over the rise of “shadow AI.”
AI Red Teaming introduces a relentless, automated vulnerability hunter that simulates real-world attackers. Instead of once-a-year assessments, AIRS runs more than 500 specialized attacks continuously, flagging weaknesses in generative AI systems long before adversaries can exploit them.
AI Model Security goes deep into the architecture of open-source and custom models, uncovering AI-native threats invisible to traditional scanners, including embedded backdoors, poisoned data inputs, and hidden malicious code. It also provides a full breakdown of a model’s architecture, datasets, licenses, and dependencies, giving enterprises the visibility needed for responsible governance and compliance.
Looking ahead, Palo Alto Networks says AIRS 2.0 is only the beginning. The company is already preparing defenses for emerging risks—from autonomous workforce systems like Cortex AgentiX to secure browsers, and even future threats such as quantum-level vulnerabilities. With a fully integrated and rapidly evolving AI security platform, it aims to help enterprises scale AI adoption with confidence and future-proof their digital ecosystems.
Prisma AIRS 2.0 is now available. More updates will be shared at Ignite: What’s Next on October 28. For further announcements, visit the Palo Alto Networks news center and follow the company on X, LinkedIn, Facebook, and Instagram.