Compare · 2 to 6 vendors

GenAI security facts side by side: compare two to six vendors

In short

Choose two to six vendors to put their published AI security facts in one table: the three approaches, thirteen checks, stated responses, network wording, compliance and analyst placement as each vendor states it. No totals or ranks are shown. The page opens with all six in alphabetical order.

As stated on each vendor's own pages, read 7 October 2026. Alphabetical order.
Item	Cato Networks	Cisco	Fortinet	Netskope	Palo Alto Networks	Zscaler
Approaches
AI controls in a SASE or SSE cloud	Described Note Cato states AI Security checks every interaction inline on its SASE cloud, with one engine and one console for users, applications and agents. Source: Cato Networks: AI security · Read 7 Oct 2026	Described Note AI Access is described as a fully integrated feature set of Cisco Secure Access SSE that inspects web traffic to identify shadow AI. Source: Cisco Secure Access: AI Access · Read 7 Oct 2026	Partly described Note Fortinet sells FortiSASE, and its AI security page describes AI app discovery and prompt blocking; which enforcement point delivers these is not stated on the pages we read. Source: Fortinet: AI security · Read 7 Oct 2026	Described Note Netskope's generative AI page describes fine-grained policies by user behaviour and data sensitivity, with controls on uploading, copying and printing within AI apps. Source: Netskope: securing generative AI · Read 7 Oct 2026	Described Note AI Access Security is listed in the Prisma SASE product family and gives visibility into shadow AI with inline data controls. Source: Palo Alto Networks: AI Access Security · Read 7 Oct 2026	Described Note AI Access Security detects and classifies AI apps and moderates prompt content on the Zero Trust Exchange. Source: Zscaler: AI Access Security · Read 7 Oct 2026
Standalone AI security layer	Described Note Protects homegrown AI apps, blocking prompt injection and jailbreaks, through the Cato cloud or the AI gateway you already run. Source: Cato Networks: AI security · Read 7 Oct 2026	Described Note AI Defense includes AI Runtime Protection, AI Model and Application Validation, AI Cloud Visibility, AI Access and AI Supply Chain Risk Management. Source: Cisco AI Defense · Read 7 Oct 2026	Described Note FortiAIGate secures and governs AI apps and agent traffic at runtime from one control point. Source: Fortinet: FortiAIGate · Read 7 Oct 2026	Described Note An AI Gateway for API traffic between custom-built apps and LLMs, and AI Guardrails as a dedicated runtime defense layer. Source: Netskope: AI Guardrails · Read 7 Oct 2026	Described Note Prisma AIRS includes an AI Gateway, AI Runtime Security, AI Red Teaming, AI Posture Management and Agent Identity Security. Source: Palo Alto Networks: Prisma AIRS · Read 7 Oct 2026	Described Note An AI Gateway routes, secures and governs every AI transaction, with inline DLP applied to every model, agent and API call. Source: Zscaler: AI security · Read 7 Oct 2026
Firewall or self-hosted AI gateway	Partly described Note Homegrown AI apps can also be protected on-prem with Outposts; self-hosted enforcement for employee AI use is not described. Source: Cato Networks: AI security · Read 7 Oct 2026	Not in our sources Note Not described on the pages we read.	Described Note FortiAIGate is described as a purpose-built FortiGate; with self-hosting, data and logs stay on your infrastructure. Source: Fortinet: FortiAIGate · Read 7 Oct 2026	Not in our sources Note Not described on the pages we read.	Not in our sources Note Not described on the pages we read.	Not in our sources Note Not described on the pages we read.
Checks
Finds shadow AI use	Stated Note Discovers shadow AI and coding agents across endpoints and auto-inventories every third-party AI tool, copilot and agent. Source: Cato Networks: AI security for end users · Read 7 Oct 2026	Stated Note Inspects web traffic to identify shadow AI, including API calls made to AI models inside AI-enhanced applications. Source: Cisco Secure Access: AI Access · Read 7 Oct 2026	Stated Note Discovers AI applications in use with their categories, use cases, risk levels and log details. Source: Fortinet: AI security · Read 7 Oct 2026	Partly stated Note Continuous discovery of remote MCP servers, clients, tools, resources and prompt requests, and an App Catalog MCP inventory; general GenAI app discovery is not detailed on the pages we read. Source: Netskope: Agentic Broker · Read 7 Oct 2026	Stated Note Visibility into shadow AI, covering GenAI applications, agents and marketplace plugins, with real-time visibility into which apps are used and by whom. Source: Palo Alto Networks: AI Access Security · Read 7 Oct 2026	Stated Note Detects and classifies thousands of AI apps, including AI embedded in popular SaaS applications; AI Asset Management auto-discovers models, agents, MCP servers and shadow AI. Source: Zscaler: AI Access Security · Read 7 Oct 2026
Inspects prompts for sensitive data	Stated Note Stops data leaks in prompts, with Monitor, Anonymize and Block as policy actions. Source: Cato Networks: AI security · Read 7 Oct 2026	Partly stated Note AI Defense names data leakage among the runtime threats it addresses; prompt-level DLP is not described on the pages we read. Source: Cisco AI Defense · Read 7 Oct 2026	Stated Note FortiAIGate checks prompts and outputs before they reach users or systems and is described as stopping data leaks. Source: Fortinet: FortiAIGate · Read 7 Oct 2026	Stated Note With Netskope One DLP, the broker applies real-time policies that identify and block sensitive information. Source: Netskope: Agentic Broker · Read 7 Oct 2026	Stated Note LLM-powered data classification augments DLP, and inline data detection blocks sensitive text- and file-based data. Source: Palo Alto Networks: AI Access Security · Read 7 Oct 2026	Stated Note Prompt and response extraction and classification, with controls to prevent sensitive data from leaving the organization. Source: Zscaler: AI Access Security · Read 7 Oct 2026
Controls AI agent and MCP traffic	Stated Note Maps agents, MCP servers and skills, and inspects agent prompts, responses and tool calls. Source: Cato Networks: AI security · Read 7 Oct 2026	Partly stated Note Secure Access registers every agent, maps it to a human owner, and quarantines compromised agents with inline runtime monitoring; MCP is not mentioned on the pages we read. Source: Cisco Secure Access · Read 7 Oct 2026	Stated Note Checks agent tool calls and catches prompt injection, jailbreaks and data leaks in MCP tool traffic. Source: Fortinet: FortiAIGate · Read 7 Oct 2026	Stated Note Agentic Broker detects and decodes MCP traffic, blocks unauthorized MCP communication to remote MCP servers and enforces least-privilege access. Source: Netskope: Agentic Broker · Read 7 Oct 2026	Partly stated Note Prisma AIRS lists Agent Identity Security and governance of AI agents at scale; MCP is not mentioned on the pages we read. Source: Palo Alto Networks: Prisma AIRS · Read 7 Oct 2026	Stated Note Secures agentic communications through MCP and A2A brokers and governs agent-to-agent traffic. Source: Zscaler: AI security · Read 7 Oct 2026
Covers coding agents	Stated Note Names enforcement for Claude Code, Cursor and Codex and stops risky coding-agent tool calls before they run. Source: Cato Networks: AI security · Read 7 Oct 2026	Not in our sources	Not in our sources	Partly stated Note The MCP inventory covers clients including ChatGPT, Cursor and VS Code; coding-agent controls beyond MCP are not described. Source: Netskope: Agentic Broker · Read 7 Oct 2026	Stated Note Prisma AIRS lists securing AI coding agents: protect code and credentials, govern agent actions. Named coding tools are not listed. Source: Palo Alto Networks: Prisma AIRS · Read 7 Oct 2026	Partly stated Note Zero trust access to AI developer tools with inline controls; named coding agents are not listed. Source: Zscaler: AI Access Security · Read 7 Oct 2026
Protects homegrown AI apps at runtime	Stated Note Blocks prompt injection and jailbreaks and red teams homegrown AI apps. Source: Cato Networks: AI security · Read 7 Oct 2026	Stated Note AI Runtime Protection places guardrails in the network to block adversarial attacks and harmful responses in real time. Source: Cisco AI Defense · Read 7 Oct 2026	Stated Note Guardrails check text, images, audio, video and code, and catch prompt injection and jailbreaks. Source: Fortinet: FortiAIGate · Read 7 Oct 2026	Stated Note AI Guardrails mitigate prompt injection and jailbreak attempts through real-time analysis, inspecting requests and responses in 29 languages. Source: Netskope: AI Guardrails · Read 7 Oct 2026	Stated Note AI Runtime Security enforces real-time safeguards to prevent manipulation, data exposure and unsafe actions. Source: Palo Alto Networks: Prisma AIRS · Read 7 Oct 2026	Stated Note The AI Gateway applies inline DLP and Zero Trust to every model, agent and API call; AI red teaming is offered. Source: Zscaler: AI security · Read 7 Oct 2026
Blocks inline in real time	Stated Note Checks every interaction inline, with fast checks in milliseconds; Block is one of three policy actions. Source: Cato Networks: AI security · Read 7 Oct 2026	Stated Note Detects and blocks malicious and undesired AI traffic using multiple enforcement points. Source: Cisco Secure Access: AI Access · Read 7 Oct 2026	Stated Note FortiAI blocks unauthorized prompts; FortiAIGate is described as stopping harmful AI content in real time. Source: Fortinet: AI security · Read 7 Oct 2026	Stated Note Blocks unauthorized MCP communication and sensitive information with real-time policies. Source: Netskope: Agentic Broker · Read 7 Oct 2026	Stated Note Inline data detection blocks sensitive data, and the page states it stops AI-generated threats in real time. Source: Palo Alto Networks: AI Access Security · Read 7 Oct 2026	Stated Note Block access by role in real time. Source: Zscaler: AI security · Read 7 Oct 2026
Coaches or warns the user	Not in our sources	Partly stated Note Can redirect requests to approved services; user coaching is not named. Source: Cisco Secure Access: AI Access · Read 7 Oct 2026	Stated Note User coaching drives correct behavior and captures artifacts for intent investigation. Source: Fortinet: AI security · Read 7 Oct 2026	Partly stated Note Fine-grained policies beyond simple allow-or-block, based on user behaviour and data sensitivity; coaching is not named. Source: Netskope: securing generative AI · Read 7 Oct 2026	Stated Note User coaching notifications when people access unsanctioned apps. Source: Palo Alto Networks: AI Access Security · Read 7 Oct 2026	Stated Note Allow, block or coach access by users or user groups. Source: Zscaler: AI Access Security · Read 7 Oct 2026
Has a watch or log-only action	Stated Note Monitor is one of three policy actions. Source: Cato Networks: AI security · Read 7 Oct 2026	Partly stated Note Inline runtime monitoring and semantic inspection are described for agents; a watch-only action for employee AI use is not named. Source: Cisco Secure Access · Read 7 Oct 2026	Stated Note Apply guardrails, monitor activity and adjust policies in real time. Source: Fortinet: AI security · Read 7 Oct 2026	Not in our sources	Not in our sources	Not in our sources
Redacts or anonymizes data in prompts	Stated Note Anonymize is a policy action; Cato states it redacts sensitive data inline. Source: Cato Networks: AI security for end users · Read 7 Oct 2026	Not in our sources	Not in our sources	Not in our sources	Not in our sources	Partly stated Note Filters sensitive data from LLMs; how data inside a prompt is masked is not described. Source: Zscaler: AI security · Read 7 Oct 2026
Can run on your own infrastructure	Partly stated Note On-prem protection with Outposts is described for homegrown AI apps, not for employee AI use. Source: Cato Networks: AI security · Read 7 Oct 2026	Not in our sources	Stated Note With self-hosting, data and logs stay on your infrastructure; FortiSASE also has a Sovereign option deployed in the customer's own data centres. Source: Fortinet: FortiAIGate · Read 7 Oct 2026	Not in our sources	Not in our sources	Not in our sources
States FedRAMP	Not in our sources	Partly stated Note Cisco Secure Access states FedRAMP authorization covering Secure Internet Access, VPNaaS and ZTNA; AI Access is not named in that scope. Source: Cisco Secure Access · Read 7 Oct 2026	Not in our sources	Partly stated Note Netskope states it achieved FedRAMP High Authorization (GovCloud); which services are in scope is not stated on the page we read. Source: Netskope: security, compliance and assurance · Read 7 Oct 2026	Partly stated Note The trust center certifications page lists FedRAMP and StateRAMP; which products each covers is not stated in the text we read. Source: Palo Alto Networks trust center: certifications · Read 7 Oct 2026	Not in our sources
Lists security attestations	Stated Note Lists SOC 2 and SOC 3, PCI-DSS Level 1 Service Provider, ISO/IEC 27001, 27017, 27018 and 27701, CSA STAR Self-Assessment, UK Cyber Essentials and Spain ENS High. Source: Cato Networks: security, compliance and privacy · Read 7 Oct 2026	Not in our sources	Partly stated Note The trust page states alignment with the ISO/IEC 27000 series and NIST 800-53; specific certifications for FortiSASE are not stated in the text we read. Source: Fortinet: trust · Read 7 Oct 2026	Stated Note Lists AICPA SOC 2, C5, CSA STAR Level II, IRAP and the Data Privacy Framework; SOC 2 and C5 reports on request. Source: Netskope: security, compliance and assurance · Read 7 Oct 2026	Stated Note The trust center lists SOC 2+, Germany C5, PCI DSS and IRAP, among others. Source: Palo Alto Networks trust center: certifications · Read 7 Oct 2026	Not in our sources
One console for AI and network policy	Stated Note Cato states one AI-native engine and one console cover users, applications and agents. Source: Cato Networks: AI security · Read 7 Oct 2026	Stated Note Single console, unified client and centralized policy management for Secure Access, with AI Access integrated into it. Source: Cisco Secure Access · Read 7 Oct 2026	Partly stated Note Unified SASE is built on a single OS, policy engine and management plane; FortiAIGate is described as a separate control point. Source: Fortinet: Unified SASE · Read 7 Oct 2026	Partly stated Note Netskope One states one engine, one client, one console and one network; how the AI products sit in that console is not described. Source: Netskope One · Read 7 Oct 2026	Partly stated Note Strata Cloud Manager with Strata Copilot manages SASE policy; Prisma AIRS is described as its own platform. Source: Palo Alto Networks: AI Access Security · Read 7 Oct 2026	Partly stated Note Zero Trust SASE Everywhere lists branch, SSE and AI security together; one console for AI and network policy is not described. Source: Zscaler: Zero Trust SASE · Read 7 Oct 2026
Responses
Watch, coach, redact, block	Cato Networks states Watch, Redact and Block; Coach is not in our sources.	Cisco states Block; Watch and Coach are partly stated; Redact is not in our sources.	Fortinet states Watch, Coach and Block; Redact is not in our sources.	Netskope states Block; Coach is partly stated; Watch and Redact are not in our sources.	Palo Alto Networks states Coach and Block; Watch and Redact are not in our sources.	Zscaler states Coach and Block; Redact is partly stated; Watch is not in our sources.
In the vendor's words
Network	A cloud-native, private global backbone spanning 100+ PoPs worldwide, with a 99.999% uptime SLA. The SLA terms are not on the page. Source: Cato Networks: global private backbone · Read 7 Oct 2026	No PoP or data-centre count on the pages we read. Source: Cisco Secure Access · Read 7 Oct 2026	A global network of 200+ PoPs on Fortinet-owned infrastructure, extended through integration with AWS, Google Cloud and Oracle Cloud. Source: Fortinet: FortiSASE · Read 7 Oct 2026	NewEdge is powered by 120+ data centers in 80+ regions, including Mainland China, with full compute and all SASE services at the edge in every location. Source: Netskope: NewEdge · Read 7 Oct 2026	A 99.999% uptime SLA on its hyperscale backbone. No PoP or location count in the Prisma Access page text. Source: Palo Alto Networks: Prisma Access · Read 7 Oct 2026	No data-centre or PoP count on the pages we read. Source: Zscaler: Zero Trust Exchange · Read 7 Oct 2026
Compliance	SOC 2 and SOC 3, PCI-DSS Level 1 Service Provider, ISO/IEC 27001, 27017, 27018, 27701 and 14001, UK Cyber Essentials, CSA STAR Self-Assessment, GDPR and Spain ENS High. FedRAMP is not listed on the page. Source: Cato Networks: security, compliance and privacy · Read 7 Oct 2026	FedRAMP authorization for Secure Internet Access, VPNaaS and ZTNA. Other certifications are not stated on the page. Source: Cisco Secure Access · Read 7 Oct 2026	PSIRT disclosure aligned with ISO/IEC 30111 and 29147; security aligned with the ISO/IEC 27000 series and NIST 800-53. FortiSASE certifications such as SOC 2 or FedRAMP are not stated in the text we read. Source: Fortinet: trust · Read 7 Oct 2026	FedRAMP High Authorization (GovCloud), AICPA SOC 2, C5, CSA STAR Level II, IRAP and the Data Privacy Framework. Source: Netskope: security, compliance and assurance · Read 7 Oct 2026	The trust center lists SOC 2+, Germany C5, PCI DSS, IRAP, FedRAMP and StateRAMP, among others; product scope is not stated in the text we read. Source: Palo Alto Networks trust center: certifications · Read 7 Oct 2026	Not stated on the pages we read. Source: Zscaler: Zero Trust SASE · Read 7 Oct 2026
Analyst placement	Named a Leader in the 2026 Gartner Magic Quadrant for SASE Platforms for the third consecutive year (Gartner, Magic Quadrant for SASE Platforms, 28 July 2026). Source: Cato Networks news: Leader in the 2026 Gartner Magic Quadrant for SASE Platforms · Read 7 Oct 2026	The SASE page says "Cisco named a SASE Leader"; the analyst firm and report name are not given in the page text. Source: Cisco: SASE solution · Read 7 Oct 2026	Recognized as a Challenger in the 2026 Gartner Magic Quadrant for SASE Platforms. Source: Fortinet: FortiSASE · Read 7 Oct 2026	A 2026 Gartner Magic Quadrant Leader for SASE Platforms and for Security Service Edge. Source: Netskope: 2026 Gartner Magic Quadrant reports for SASE Platforms and SSE · Read 7 Oct 2026	The Prisma SASE page says "A 4X Gartner Magic Quadrant Leader for SASE Platforms". Its Prisma Access page uses a different count, so we cite the Prisma SASE page only. Source: Palo Alto Networks: Prisma SASE · Read 7 Oct 2026	Recognized as a Leader in the 2026 Gartner Magic Quadrant reports for SASE and SSE. Source: Zscaler: Zero Trust SASE · Read 7 Oct 2026
Pricing	Not published. The address catonetworks.com/pricing/ returned a not-found page on 7 October 2026. Source: Cato Networks /pricing/ address (returned a not-found page) · Read 7 Oct 2026	Not published on the pages we read. Source: Cisco Secure Access · Read 7 Oct 2026	Not published. Fortinet mentions flexible licensing packages and links an ordering guide, which we did not open. Source: Fortinet: FortiSASE · Read 7 Oct 2026	Not published on the pages we read. Source: Netskope One · Read 7 Oct 2026	Not published. The Prisma Access page implies per-mobile-user licensing. Source: Palo Alto Networks: Prisma Access · Read 7 Oct 2026	Not published on the pages we read. Source: Zscaler: Zero Trust SASE · Read 7 Oct 2026

Checks every selected vendor states: Protects homegrown AI apps at runtime and Blocks inline in real time.

Checks none of the selected vendors states: States FedRAMP.

Questions

Why are there no totals?

Because a total would hide the check that matters to you. Use the quiz to weight checks by your own answers.

Can I share a comparison?

Yes. The selected vendors are kept in the address bar.

Why do network figures look different?

Each vendor states its network on its own basis, such as PoPs, data centers or regions. We quote each as written and do not compare them.