Agentic Innovation Meets Enterprise Control

Today, we are introducing new security innovations that help enterprises put AI agents to work with greater control over sensitive data, agent actions and AI costs.

As agents move from answering questions to writing code, using tools and executing workflows, they gain access to more of the business. Enterprises need protection that keeps pace with those capabilities so teams can adopt AI confidently while maintaining oversight.

Our latest innovations address three priorities:

• Secure frontier AI adoption with runtime protection integrated directly into Claude Enterprise.
• Protect the broader AI supply chain by expanding scanning beyond models to agent skills and other components.
• Enforce control as agents act with new AI Gateway capabilities for tool security and intelligent routing.

Together, these advances give enterprises more precise control over how agents access data, use tools and consume resources, helping turn agentic innovation into business value.

Bringing Enterprise-Grade Security to the AI Frontier

Tools like Claude are increasingly used for coding and business workflows. Integrating security directly into these environments allows organizations to enforce policies and protect sensitive data without adding friction to everyday work.

Prisma AIRS integrates directly with Claude Enterprise through inference hooks, bringing runtime protection to Claude Code, Claude.ai and Claude Cowork. This helps your teams:

• Accelerate Claude adoption with clear security boundaries. Apply runtime guardrails before model inference and enforce policies on what interactions are allowed.
• Protect sensitive data without duplicating security tasks. Extend existing enterprise DLP policies to help keep PII, intellectual property and source code from being exposed without creating and maintaining separate rule sets.
• Reduce the risk of AI breaches. Block prompt injections, jailbreaks and toxic inputs, protecting the workflows your business increasingly depends on.

For a deeper look, watch Spencer Thellmann of Palo Alto Networks and Dor Fledel of Anthropic in our fireside chat on securing frontier AI in production.

Securing the AI Supply Chain Beyond the Model

Protecting AI interactions is only part of the equation. Enterprises must also secure the components that power AI applications and agents, including models, agent skills, plugins, MCP servers, and open-source dependencies. Unified visibility and risk assessment, from development through runtime, help security teams enable developers to innovate with confidence.

These components accelerate development, but unscanned code can expose data and credentials. Fragmented security tools make it harder to spot those risks and maintain visibility after deployment.

With 3M+ third-party artifacts pre-scanned, Prisma AIRS AI Supply Chain Security helps teams adopt AI with confidence. It scans open-source models, validates agent skills and blocks malicious code before deployment. It also helps prevent threats from reaching production.

Real-Time Agent Governance

Trusted components are the foundation. Enterprises also need control over how agents use models, tools and data in production. As adoption spreads across teams, the Prisma AIRS AI Gateway centralizes policy enforcement and usage oversight to help reduce inconsistent controls and unpredictable costs.

Since launch, more than 20 major gateway updates have expanded model and provider support (including support for decision models), added customer-requested controls and deepened integrations across Palo Alto Networks helping enterprises adopt new AI capabilities while maintaining oversight.

Beyond these core advances, two additional enhancements help teams protect AI environments with less operational complexity:

MCP Guardrails: Protect Data and Production Systems

Give agents access to enterprise tools with controls that help prevent data exposure and unauthorized actions. AI Gateway inspects MCP traffic in both directions, blocks concealed instructions in tool inputs, redacts sensitive data from outputs and enforces access policies bringing prompts and tool calls under one security framework.

Signals & Smart Routing: Make AI Spend Go Further

Balance AI cost, quality and speed without constantly rewriting routing rules. Signals assess each request’s intent and complexity, enabling Smart Routing to select the best-fit model for your goals. Real-time analytics track KPIs like savings, latency, and cache efficiency, allowing administrators to automate optimization by simply specifying target outcomes.

Shape What’s Next with Granular AI Controls

From securing Claude and scanning the broader AI supply chain to strengthening gateway governance, these Prisma AIRS innovations help enterprises get more granular AI security by protecting data, controlling agent actions and simplifying security operations.

Give your teams room to innovate, backed by clear controls over what agents access, consume and do.

Watch the latest Prisma AIRS Innovation Showcase to see these advancements in action.

Learn how to secure AI coding agents across your enterprise.