Island is the most credible architectural challenge to proxy-based SASE for AI-era workloads in 2026. Its flagship product is a Chromium fork — primarily deployed as a full replacement browser, with Zero Trust enforcement wired into the browser core rather than bolted on as a plugin. That said, Island's device story is broader than "browser fork" implies: native iOS and Android apps have shipped since 2022, and March 2026 added Island Extension, a lightweight deployment mode that layers Island's policy engine onto non-Island (third-party) browsers for organizations not ready for a full browser swap. In March 2026, Island also launched a full SASE stack — SWG, ZTNA, CASB, RBI, and DLP delivered through the browser layer without SSL break-and-inspect — alongside Island Desktop, which extends Island's policy and DLP enforcement to native thick-client and desktop applications. Up to 90% of enterprise sessions are browser-initiated, and for all of those sessions Island enforces policy with pre-encryption visibility — no TLS decryption required because the browser sees plaintext before it encrypts outbound requests. Zero customer churn. $4.8B valuation, $730M raised (Series E, Coatue-led, March 2025).
Pillar scope: ZTNA and SSE (browser-applicable criteria). SD-WAN, AIOps, and Sovereignty are null by design — Island does not compete in those domains. Non-browser traffic (SSH, RDP thick clients on non-browser paths, SD-WAN) still requires a Big Six SASE platform alongside Island.
Primary positioning: Island is not a Big Six replacement for organizations with complex non-browser workloads, SD-WAN requirements, or network security needs. It is the right primary answer for organizations where the majority of work happens in a browser, where TLS decryption creates compliance or operational complexity, and where AI-era workloads (browser-based SaaS, internal AI tools) are the primary security surface.
The Pre-Encryption Advantage
Every proxy-based SASE vendor — Netskope, Zscaler, Cloudflare, Cato, Palo Alto — intercepts traffic at the network layer after the browser has encrypted it. This requires TLS break-and-inspect: the proxy terminates the TLS session, decrypts the traffic, inspects it, re-encrypts it, and forwards it. This works, but it creates friction: legal questions about privilege and confidentiality, certificate deployment complexity, performance overhead at inspection PoPs, and application compatibility issues from broken certificate chains.
Island eliminates this for browser traffic by operating at the browser layer itself. Island calls this the Perfect Packet Architecture — the Island browser sees every keystroke, every paste, every upload, every download, every form submission — before it's encrypted. DLP inspection, clipboard control, download restrictions, and data governance all execute on plaintext without any network-layer decryption. For the roughly 90% of enterprise sessions that are browser-initiated, this means Island can enforce the full SSE policy set without a proxy in the path. Island's network layer — rebranded from Perfect Packet to Island Network Services — now also markets DNS-level device controls aimed at IoT and OT devices; this is a lightweight complement to the browser-layer story, not a substitute for SD-WAN or full network security.
Correction (2026-07-10 Codex review): Island runs on any OS that supports Chromium — Windows, macOS, Linux, and ChromeOS — and, contrary to an earlier version of this document, Island is not desktop/Chromium-OS only. Island has shipped native iOS and Android apps since 2022, confirmed current in the App Store and Google Play as of this review. Feature depth is strongest on desktop; mobile parity for the full DLP and browser-isolation policy set should be verified per deployment before assuming desktop-equivalent coverage.
- Clipboard control at the OS layer — enforced before data reaches the OS clipboard API, not possible from an extension
- Download control with server-side watermarking — Island's browser process intercepts downloads before they touch the filesystem
- Screenshot prevention at the GPU composition layer — captures at the OS level are intercepted before rendering
- Print control at the print spooler level — print jobs can be blocked, watermarked, or audited
- Full JavaScript engine visibility — the browser's own JS engine is accessible for form field monitoring, paste inspection, and input control
- Non-browser protocol tunneling (SSH, native RDP, arbitrary UDP) still requires a separate SASE platform — this is a network-layer gap, not a device-platform gap. March 2026's Island Desktop extends Island's DLP/policy enforcement to native thick-client and desktop apps, narrowing but not closing this gap: it addresses app-level policy control, not protocol-level tunneling. Island's own materials are internally inconsistent on how far that app-level control actually reaches — see the DLP section below for the specific conflict between Island's Endpoint Services and Data Services pages; this is a genuinely unresolved scope question, not just an unverified claim.
- Requires adoption of a new browser (or, as of March 2026, Island Extension on a third-party browser) — change management overhead; user acceptance varies by organization culture
- Full SASE stack is new (March 2026) — evaluate production maturity at scale before treating as equivalent to an established Big Six platform
- No SD-WAN, no AIOps, no sovereignty certifications — Island is the enforcement layer, not the full stack for organizations with these requirements
ZTNA Analysis
Island's ZTNA enforces access policy at the browser session layer. Users authenticate via the organization's IdP (Okta, Microsoft Entra ID, Ping) before browser sessions are established. Per-application access rules are enforced by the Island browser — different policy applies to Salesforce than to SharePoint, within the same browser session. Device posture is integrated via CrowdStrike, SentinelOne, and Intune — posture signals are evaluated before and during sessions.
Island's ZTNA story for unmanaged devices is architecturally strong. A contractor on a personal laptop installs Island and authenticates via the IdP. The session is fully governed — DLP, clipboard restrictions, download controls, and audit logging all apply — without requiring any corporate software beyond the browser itself. The device remains unmanaged; the session is fully controlled. This is differentiated from agentless ZTNA approaches (like Cloudflare's browser rendering) because Island governs the client-side session, not just the server-side presentation of the app.
Each application in Island can have its own policy: allow upload to Salesforce but restrict to specific file types; allow SharePoint read access but prevent download to personal devices; allow internal AI tool access but log all prompts and responses. These controls execute inside the browser session without requiring the user to context-switch or experience proxy-related latency. Policy is applied to the exact action — paste, upload, download, print, screenshot — not to the entire network connection.
▲ Strengths
Full IdP integration for identity-based access. Per-application session controls at action granularity (paste, upload, print, screenshot). Strong unmanaged device / contractor access story. Pre-encryption policy enforcement eliminates certificate chain issues. Zero churn — very high customer satisfaction.
▼ Watch Areas
Protocol-level ZTNA for thick-client apps (native SSH, RDP, thick ERP tunneling) still requires a separate SASE platform. March 2026's Island Desktop extends Island's policy/DLP enforcement to native thick-client and desktop apps, but this is app-level policy control, not protocol-level tunneling — and Island's own Endpoint Services vs. Data Services materials give conflicting scope claims for how far that control reaches (see SSE section below); genuinely unresolved, not just unverified. DEM for user experience diagnostics across non-browser paths is outside Island's scope. Behavioral anomaly detection beyond session-level controls is less mature than Zscaler ZDX or Palo Alto ADEM.
SSE Analysis
Island's SWG enforces URL filtering and web access policy within the browser. Because Island operates at the browser layer, web filtering applies to every tab and every URL navigation without requiring any traffic to flow through a cloud PoP. For sanctioned and unsanctioned web destinations, Island can block, allow, or isolate at the browser level — with full audit logging of user actions within each web session.
DLP is Island's structural SSE differentiator. Netskope's DLP applies at the network inspection layer after TLS decryption. Island's DLP applies at the browser layer on plaintext before encryption. For regulated industries where TLS decryption creates legal or operational complexity (attorney-client privilege in legal, physician-patient confidentiality in healthcare, financial data confidentiality regulations), Island provides equivalent DLP enforcement without the decryption requirement. The DLP capabilities cover clipboard inspection, upload scanning, download content analysis, form field monitoring, and OCR on copy-paste operations.
Refined finding (2026-07-11 review) — Island Desktop DLP parity is genuinely unresolved, not just unverified. Island's own published materials give two different scope claims for Island Desktop that don't obviously reconcile. Island's Endpoint Services page describes Island Desktop narrowly — device control, device query, and device guard, covering visibility, threat detection, and remote support access, with no mention of clipboard, screenshot, or DLP-style controls. Island's Data Services page, by contrast, describes "Last-Mile Controls" (copy/paste, upload/download, print, screenshot) as applying "across browser, desktop, cloud, and network" through a single policy engine — which implies Island Desktop does carry the same DLP action-level enforcement as the browser. No independent source resolves which framing is the operative one, and the two pages are not obviously describing the same underlying capability at different levels of detail — they read as genuinely inconsistent. This Codex takes no position on which is correct; the legacy_app_support score remains at 2 to reflect that Island Desktop is a real but only partially specified capability, not full DLP parity with the browser.
Island's CASB governs SaaS access at the browser session layer. Per-application controls cover the full activity spectrum: allow Salesforce read/write but restrict download and export; allow Google Drive upload but restrict to approved file types; enforce watermarking on file downloads from SharePoint. Shadow IT discovery covers all web-accessible SaaS apps — Island can identify unsanctioned app usage and apply risk scoring without network-layer inspection. API-mode CASB for data at rest (scanning existing SharePoint, OneDrive, or Salesforce data) is outside Island's browser-layer scope and requires a separate CASB platform for comprehensive coverage.
Island's architecture makes traditional RBI unnecessary for the Island browser itself — the browser already enforces session isolation natively. For external untrusted sites accessed through Island, Island can apply additional isolation modes that restrict what content can be interacted with, rendered, or copied, going beyond what traditional pixel-push or DOM-reconstruction RBI provides. For unmanaged devices accessing Island in a browser-within-browser configuration, Island delivers comparable isolation to traditional RBI with lower latency overhead.
Island's pre-encryption visibility is directly relevant to GenAI governance. AI tool prompts — the data sensitivity problem driving GenAI DLP investment across the Big Six — are plaintext inputs in the browser before they're encrypted and sent to the AI service. Island can inspect, classify, block, or redact sensitive content in AI prompts without any network-layer decryption. DLP policies that apply to file uploads and clipboard paste also apply to AI prompt submissions, with the same classification depth available for text inputs as for file content. This is architecturally superior to proxy-based GenAI DLP for browser-based AI tool access, which requires decrypting the HTTPS session to inspect prompt content.
▲ Strengths
Full SASE stack (SWG, ZTNA, CASB, RBI, DLP) launched March 2026. Pre-encryption DLP — no TLS break-and-inspect for browser traffic. Best GenAI governance for browser-based AI tools. Per-application session controls at action granularity. Up to 90% of sessions direct without backhaul — latency advantage over proxy-based SSE. Zero customer churn.
▼ Watch Areas
Network-layer, non-browser traffic still requires a separate SASE platform for SSE coverage. March 2026's Island Desktop narrows — but does not close — the thick-client gap by extending Island's policy/DLP enforcement to native desktop apps at the app-policy level (not protocol-level tunneling); Island's own Endpoint Services and Data Services pages give conflicting scope claims on parity with browser-native DLP — genuinely unresolved, not just unverified. API-mode CASB (data at rest scanning) outside scope. Full SASE stack is March 2026 GA — evaluate production maturity at enterprise scale before treating as equivalent to established Big Six SSE. IPS/threat intelligence at the network layer is outside scope. No private PoP infrastructure for performance SLAs.
Out of Scope — SD-WAN, AIOps, Sovereignty
Island does not provide WAN connectivity, branch CPE, path steering, or SD-WAN services. For organizations with SD-WAN requirements, Island must be paired with a Big Six SASE or dedicated SD-WAN platform.
Island does not provide enterprise UEBA, GenAI policy authoring, cross-product event correlation, or network path diagnostics. Session audit logging and DLP event reporting are available; enterprise AIOps requires a separate platform.
Island does not hold FedRAMP, BSI C5, IRAP, or BYOK as of Q2 2026. For regulated industries requiring these certifications, Island must be paired with a sovereignty-compliant Big Six SASE platform. Verify current certification status with Island directly.
Architecture Patterns — Island + Big Six
Island is most architecturally powerful as the browser enforcement layer in a deployment that also includes a Big Six SASE for non-browser traffic. The right pairing depends on the organization's primary non-browser requirements:
| Primary Non-Browser Need | Recommended Big Six Pairing | Island's Role |
|---|---|---|
| SD-WAN + ZTNA convergence | Cato Networks | Island handles browser-based SaaS and AI tool governance without TLS decryption; Cato handles SD-WAN, non-browser ZTNA, and network security. |
| Regulated DLP for non-browser traffic | Netskope | Island handles browser sessions pre-encryption; Netskope handles API-mode CASB data at rest scanning, non-browser DLP, and network-level SSE. The two DLP engines operate on different traffic surfaces without redundancy. |
| Threat intelligence + hybrid NGFW estate | Palo Alto Prisma Access | Island handles browser layer enforcement; Palo Alto handles threat prevention (WildFire/App-ID), physical NGFW policy continuity, and non-browser security. Note: Palo Alto Prisma Access Browser (PAB) competes directly with Island — evaluate both before committing. |
| Per-app ZTNA for thick-client apps | Zscaler ZPA | Island handles browser-based app access; ZPA handles thick-client private app access via inside-out connector model. ZDX provides DEM for the full user experience across both surfaces. |
Competitive Note — Palo Alto Prisma Access Browser
Palo Alto's Prisma Access Browser (PAB), from the 2023 Talon Security acquisition, is Island's most direct Big Six competitor. Both are Chromium forks targeting enterprise browser enforcement. The key differences as of Q2 2026:
| Dimension | Island | Palo Alto PAB |
|---|---|---|
| SASE integration | Standalone + full SASE stack (March 2026) | Integrated with Prisma Access + SCM policy plane — shares App-ID, DLP, and access policy with the broader PA stack |
| Policy plane | Island-native policy engine | SCM — same policy plane as Prisma Access cloud SASE and physical NGFWs |
| Maturity | More mature browser product, broader market traction, zero churn | PAB + SCM integration still maturing as of Q1 2026 — verify current integration scope before positioning |
| Best for | Organizations without existing Palo Alto investment; organizations wanting browser-layer SASE without full PA stack commitment | Organizations already running Prisma Access and SCM who want browser enforcement unified with their existing PA policy plane |
For organizations already running Palo Alto, PAB should be evaluated alongside Island before making a standalone enterprise browser decision. For organizations without Palo Alto, Island is the stronger current choice on product maturity and market traction.
Market Position — Gartner Secure Enterprise Browsers
Gartner tracks Island in the Secure Enterprise Browsers market (4.8★ Peer Insights, Q1 2026). Gartner forecasts 25% of organizations will deploy at least one secure enterprise browser by 2028, up from approximately 10% today — a market Island is positioned to lead as browser-layer enforcement gains acceptance across regulated and AI-heavy organizations.
Persona Fit Summary
| Persona | Island Fit | Primary Reason | Watch |
|---|---|---|---|
| Lean IT SMB–Mid-market | VIABLE | Island's full SASE stack (March 2026) may be sufficient for lean IT teams with browser-heavy workloads and minimal non-browser security requirements. Simpler than a Big Six proxy-based deployment for browser-first organizations. | No SD-WAN, no network security for non-browser traffic. Evaluate whether 90%+ browser session assumption holds for the actual workload profile. |
| Global Security Ops Large Enterprise | COMPLEMENT | Island complements a Big Six SASE by handling browser-layer enforcement without TLS decryption — reduces proxy inspection scope for managed-device SaaS sessions. Best positioned as a layer alongside Palo Alto or Zscaler, not as a replacement. | Full SASE stack is March 2026 GA — validate at scale. Change management for browser adoption. Requires integration planning with existing security tooling. |
| Data-First / Regulated Finance · Healthcare · Legal | PRIMARY USE CASE | Pre-encryption DLP without TLS break-and-inspect is the architecturally correct answer for industries where TLS decryption creates legal privilege or regulatory complexity. Island's browser-layer DLP enforces the same controls without the decryption requirement. | Sovereignty certifications (FedRAMP, BSI C5) not held — must be paired with a certified Big Six SASE for regulated environments with certification requirements. API-mode CASB requires a separate platform for data-at-rest coverage. |
| Platform / Network Architect 500–5,000 employees | NOT RECOMMENDED STANDALONE | Network architects evaluating SD-WAN + ZTNA + SSE convergence need a platform with network-layer capabilities that Island doesn't provide. Island as a browser enforcement layer alongside Cato or Cloudflare is a valid architecture; Island as the primary SASE is not for this persona. | — |
Changelog
| Date | Version | Change |
|---|---|---|
| 2026-07-22 | v1.5 | Island's new platform claims (DEM, Island Private Access, SaaS API) and the Desktop DLP-parity conflict formally accepted as standing watch items, not active pending research — a further web search pass found only more vendor-authored material (plus a tangential real fact: Microsoft Purview DLP integration reaching GA late June–July 2026), no independent corroboration. Decision made per user request; scores held (dem=2, legacy_app_support=2, casb_dual_mode=3). |
| 2026-07-22 | v1.4 | Codex-wide accuracy review pass: Island's March 17, 2026 "Island Enterprise Platform" relaunch (Endpoint/Data/Network/Identity/AI/Productivity Services) appears to have expanded real capabilities beyond what this Codex reflects — possible upward revisions flagged for ztna.dem (native DEX component), ztna.legacy_app_support (new "Island Private Access" protocol-level tunnel), and sse.casb_dual_mode (new native "SaaS API" service, contradicting the prior "no API-mode CASB" note). Not applied this pass — only vendor-authored site copy was found, and this Codex's evidentiary standard calls for independent corroboration before a score change; scheduling a fuller verification pass. Island Desktop DLP-parity conflict (v1.3) is somewhat narrowed by Island's current site (standalone Desktop now claims native clipboard/print/USB/screenshot DLP, with browser-pairing needed for full download/upload protection) but not fully resolved — score held at 2. |
| 2026-07-11 | v1.3 | Refined Island Desktop DLP-parity note — Island's own Endpoint Services and Data Services pages give conflicting scope claims; genuinely unresolved, not just unverified. Score held at 2. |
| 2026-07-10 | v1.2 | Codex review pass: CORRECTED factual error — Island has supported iOS/Android since 2022 (os_coverage 4→5 in scores.json). Updated legacy_app_support framing for March 2026 Island Desktop launch (1→2). Added Island Extension and Perfect Packet/Network Services mentions. Also fixed a pre-existing orphaned closing <div> (Market Position section) found during div-balance audit. |
| 2026-04-20 | v1.1 | Qualified "zero customer churn" claim with Island-reported attribution; added Perfect Packet Architecture terminology; added Fortune 100 VDI displacement proof point; added Gartner Secure Enterprise Browsers market context and 2028 forecast. |
| 2026-04-19 | v1.0 | Initial working document created under v2.0 Codex structure. Upgraded from ZTNA-only stub to full ZTNA + SSE coverage following Island's March 2026 full SASE stack launch (SWG, ZTNA, CASB, RBI, DLP). Content expanded from sase_emerging.html contextual analysis. $4.8B valuation, $730M raised, zero churn noted. PAB competitive comparison added. |