Last updated: November 3, 2026. Testing track: Documentation Reviewed.
Zscaler applies its Zero Trust Exchange proxy to generative AI, with allow, caution, block and isolate controls plus inline DLP for AI apps, and since January 2026 a broader AI security suite. The SPLX acquisition added automated AI red teaming, prompt hardening and AI asset discovery for custom AI apps. It is strongest for organizations already routing traffic through Zscaler.
Ownership: Public company (Nasdaq: ZS). Acquired SPLX, announced November 3-4, 2025 with the deal described as completed; financial terms not disclosed (Help Net Security). Also announced acquiring Symmetry Systems in May 2026 for a reported $175 million (SiliconANGLE).
Score: 7.4 / 10.
Initial review November 3, 2026.
This review is based on Zscaler’s public documentation, product pages, pricing pages and press coverage as of November 2026. We have not had lab access or a vendor demo.
Scores use the same published rubric and formula as every review on this site. Sources are listed at the end of the review.
Coverage breadth
Detection accuracy
Deployment friction
Policy & control depth
Framework alignment
Support & documentation
20%
20%
15%
15%
10%
10%
10%
9
8
6
8
7
4
8
Covers workforce GenAI, custom apps, agents, MCP, endpoints and SaaS third-party apps, though some pieces are recent announcements.
Published detector list and large DLP dictionary set, with no independent benchmarks.
Requires Zscaler traffic forwarding and SSL inspection for workforce controls; API option for app teams.
Allow, caution, block and isolate modes with per-app DLP, but fewer documented redact or justification options.
Red teaming mapped to MITRE ATLAS, NIST and EU AI Act on Zscaler’s page, without detailed docs.
Quote only.
Large help portal and partner guides; newer AI features lack detailed public docs.
Score
9
Notes
Covers workforce GenAI, custom apps, agents, MCP, endpoints and SaaS third-party apps, though some pieces are recent announcements.
Score
8
Notes
Published detector list and large DLP dictionary set, with no independent benchmarks.
Score
6
Notes
Requires Zscaler traffic forwarding and SSL inspection for workforce controls; API option for app teams.
Score
8
Notes
Allow, caution, block and isolate modes with per-app DLP, but fewer documented redact or justification options.
Score
7
Notes
Red teaming mapped to MITRE ATLAS, NIST and EU AI Act on Zscaler’s page, without detailed docs.
Score
4
Notes
Quote only.
Score
8
Notes
Large help portal and partner guides; newer AI features lack detailed public docs.
Offers four GenAI access modes (allow, caution, block, isolate), including browser isolation that blocks downloads of AI output, per Zscaler.
SPLX adds automated red teaming with 5,000+ attack simulations and discovery of models, RAG systems and MCP servers, per Help Net Security.
Covers OAuth-connected third-party apps in Microsoft 365 and Google with revocation, per Zscaler's third-party app security page.
AI Guard publishes a detector list (prompt injection, jailbreak, secrets, PII, toxicity and more) and integrates with NVIDIA NeMo Guardrails.
Workforce controls depend on routing traffic through Zscaler with SSL inspection, a heavier deployment than browser extensions.
Pricing for AI Protect is not published.
Several 2026 capabilities (AI Broker, Endpoint AI Security) were announced without GA dates or pricing, per SiliconANGLE (June 2026). No independent detection benchmarks found for AI Guard or red teaming. Product naming has shifted quickly (AI Security Suite, AI Protect, AI Guard, SPLX), which complicates evaluation.
Enterprises that already run Zscaler Internet Access and want GenAI controls, DLP and browser isolation inside their existing zero trust proxy, and that also build AI applications needing red teaming and prompt hardening from SPLX technology. Strong for regulated industries that want an isolate option for AI apps.
Organizations without Zscaler that need fast, low-friction governance of employee AI use may prefer browser-extension tools such as Harmonic, Nudge or Portal26. Teams wanting transparent public pricing or a standalone developer guardrail API may look at Lakera.
Not published. Quote-based through Zscaler and partners; no AI Protect list prices found.
See how it compares in our Best AI DLP Tools 2026 ranking. The poor-fit section above names alternatives for buyers it does not suit, and our methodology explains how every product is scored.
If Zscaler grants lab access, we would run the following scenarios and upgrade this review to Lab Tested.
The standard sensitive-data set (PII, PHI, payment data, secrets, source code) sent through the product's protected AI tools.
Inventory accuracy against a known set of sanctioned and unsanctioned AI tools.
Block, warn, redact and allow behaviors against a configured policy.
What is logged, what is not, and retention behavior.
Microsoft Entra ID and Okta where supported.
Time from access to enforced policy in a representative environment.
Added latency on standard prompt sizes.
Based on public documentation. “Not documented” means we could not find it in public material, not that the product lacks it.
| Capability | Zscaler |
|---|---|
| Inline prompt DLP | Yes. Inline SSL inspection and DLP on GenAI prompts and uploads; multi-turn prompt inspection in AI Protect. |
| Shadow AI discovery | Yes. AI Asset Management covers 2,900+ apps plus agents, models and MCP servers in public cloud. |
| Blocking unapproved AI tools | Yes. Block the AI and ML Applications category or specific apps, or allow with caution. |
| File upload inspection | Yes. DLP policies inspect uploads of sensitive data such as source code and PII into AI apps. |
| Secure AI workspace (sanctioned chat) | Partial. Browser isolation lets users reach AI apps in an isolated session without downloads; not a vendor chat interface. |
| Built-in AI vendor risk scoring | Partial. SaaS supply chain security profiles third-party apps; AI app-specific risk scores not documented in sources opened. |
| Microsoft 365 connected AI app review and control | Yes. SaaS supply chain security scans Microsoft 365 for third-party app connections and can revoke them; AI-specific workflow not described. |
| Google Workspace connected AI app review and control | Yes. Google is listed as a supported platform for third-party app discovery and revocation. |
| Governance scorecard, attestation and board reporting | Partial. Compliance Heat Map and framework-mapped red team results; no board report or attestation documented. |
| AI agent and MCP security | Yes. MCP red teaming, MCP gateway, and AI Broker for MCP and A2A traffic with an agent registry (2026 announcements). |
| Prompt injection defense | Yes. AI Guard detects prompt injection and jailbreaks; standalone Prompt Hardening Service. |
| AI red teaming | Yes. SPLX automated AI red teaming with 5,000+ attack scenarios. |
| SIEM integration | Yes. Nanolog Streaming Service streams Zscaler logs to SIEMs such as Splunk and QRadar. |
Inline proxy via Zscaler Internet Access (Client Connector or other traffic forwarding) with SSL inspection, browser isolation, API integration (AI Guard detection API, including NVIDIA NeMo Guardrails integration), SaaS API connectors, and AI Gateway. Rollout time not published; assumes Zscaler traffic forwarding is in place for workforce controls.
Public GenAI web apps (AI Asset Management cites 2,900+ apps; prompt extraction across 250+ GenAI apps as of June 2026), Anthropic and OpenAI enterprise compliance APIs, custom LLM apps (AI Guard API, red teaming, prompt hardening), AI agents and MCP servers in AWS, Azure and GCP, agent-to-agent and MCP traffic via AI Broker (announced June 2026), endpoint browsers, extensions and local AI tools (Endpoint AI Security, announced June 2026), and third-party SaaS app connections via SaaS supply chain security.
Yes. Zscaler announced it had acquired SPLX in early November 2025; terms were not disclosed. SPLX adds AI red teaming, prompt hardening and AI asset discovery.
Yes. Zscaler can block, allow with a caution page, or isolate ChatGPT and other AI apps, and apply DLP to prompts and uploads.
Zscaler’s AI security suite launched in January 2026, covering AI Asset Management, Secure Access to AI and Secure AI Infrastructure and Apps, expanded at Zenith Live in June 2026.
Zscaler does not publish pricing for AI Protect; it is sold by quote.