Three approaches
GenAI control approaches: where can AI security sit?
In short
A GenAI control can sit in the SASE or SSE cloud that already carries your users' traffic, in a standalone AI security layer in front of models, apps and agents, or on a firewall or AI gateway you run yourself. Most companies end up with more than one. The quiz tells you where to start; this page explains what each position sees.
- SSE · Security cloud (SASE or SSE)
- AIG · AI gateway and guardrails
- FWG · Firewall or self-hosted gateway
- Filled square: described on the vendor pages we read
- Hatched square: partly described (a narrower scope, or the enforcement point is not stated)
- Dashed square: not in our sources (it may exist; we did not find it)
- SSE
AI controls in a SASE or SSE cloud
Adds GenAI discovery, prompt inspection and blocking to the security cloud that already carries users' web and SaaS traffic.
5 of 6 vendors describe it, 1 partly
Described: Cato Networks, Cisco, Netskope, Palo Alto Networks and Zscaler
Partly: Fortinet
Read about AI controls in a SASE or SSE cloud - AIG
Standalone AI security layer
A dedicated AI gateway or runtime guardrail layer in front of models, homegrown AI apps and agents, managed as its own control point.
6 of 6 vendors describe it, 0 partly
Described: Cato Networks, Cisco, Fortinet, Netskope, Palo Alto Networks and Zscaler
Partly: none
Read about Standalone AI security layer - FWG
Firewall or self-hosted AI gateway
Enforcement on a firewall or AI gateway you run yourself, so AI traffic, data and logs stay on infrastructure you control.
1 of 6 vendors describe it, 1 partly
Described: Fortinet
Partly: Cato Networks
Read about Firewall or self-hosted AI gateway
Which vendor describes which approach
| Vendor | AI controls in a SASE or SSE cloud | Standalone AI security layer | Firewall or self-hosted AI gateway |
|---|---|---|---|
| Cato Networks | Described | Described | Partly described |
| Cisco | Described | Described | Not in our sources |
| Fortinet | Partly described | Described | Described |
| Netskope | Described | Described | Not in our sources |
| Palo Alto Networks | Described | Described | Not in our sources |
| Zscaler | Described | Described | Not in our sources |
Cato Networks: coverage notes
- AI controls in a SASE or SSE cloud: 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
- Standalone AI security layer: 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
- Firewall or self-hosted AI gateway: 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
Cisco: coverage notes
- AI controls in a SASE or SSE cloud: 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
- Standalone AI security layer: 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
- Firewall or self-hosted AI gateway: Not described on the pages we read.
Fortinet: coverage notes
- AI controls in a SASE or SSE cloud: 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
- Standalone AI security layer: FortiAIGate secures and governs AI apps and agent traffic at runtime from one control point.
Source: Fortinet: FortiAIGate · Read 7 Oct 2026
- Firewall or self-hosted AI gateway: 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
Netskope: coverage notes
- AI controls in a SASE or SSE cloud: 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
- Standalone AI security layer: 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
- Firewall or self-hosted AI gateway: Not described on the pages we read.
Palo Alto Networks: coverage notes
- AI controls in a SASE or SSE cloud: 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
- Standalone AI security layer: 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
- Firewall or self-hosted AI gateway: Not described on the pages we read.
Zscaler: coverage notes
- AI controls in a SASE or SSE cloud: 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: 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: Not described on the pages we read.
Questions
Can I use more than one approach?
Yes, and most large companies do. A SASE or SSE cloud covers employees using AI in the browser and SaaS, while a gateway or guardrail layer covers AI apps you build and agents calling models through APIs.
Which approach catches the most?
It depends on where the traffic flows. A control only sees what passes through it: the security cloud sees user traffic it carries, a gateway sees calls routed to it, and a self-hosted gateway sees what you send through your own infrastructure.
Is a firewall enough for AI security?
A firewall you run can host AI controls if the vendor describes them, as Fortinet does with FortiAIGate. Traffic from remote users that never passes your sites will not reach it.