Track Netskope with a free Huntertech account. Get outage alerts, watchlists, and cross-vendor monitoring in one place.
Netskope updates
Release notes and changelog entries from the vendor feed. Refine with filters, then choose All, Recent to focus the list.
Product updates
All updates
Here is the list of the new features and enhancements.
Increased SWAP Size
Increased the SWAP space to align with general recommendations and minimize memory performance pressure on the system.
Policy Support for HTTP Redirect
Netskope now supports a native HTTP Redirect policy action. This allows administrators to automatically route users to a new URL (e.g., a corporate app instance) using standard HTTP 307 headers, replacing the need for manually doing so via user notification links. To learn more, view Policy Support for HTTP Redirect.
Here is the list of fixed issues in this release.
| Issue | Description |
|---|---|
1031565 | After upgrading to appliance version 135.0.0, a timing issue during appliance startup could cause the URL database updater service to fail, resulting in the data plane (DPOP) stopping traffic processing. This occurred when an internal directory was created with incorrect permissions due to the order in which system components started up. This issue is now fixed. The required directories are now created with the correct permissions during initialization, ensuring the URL database updater starts reliably regardless of component startup order. |
1030520 | Fixed an issue where certain sites became unreachable due to an LFS configuration mismatch when LFS was enabled on the global tenant but disabled on the appliance. |
994372 | A high-severity "Diagnostic Agent Health" alarm appeared on the tenant dashboard roughly every 6 hours and cleared on its own after about 5 minutes. The diagnostic agent was healthy throughout β no data was lost and no functionality was affected. The alarm was a false positive. This issue is now fixed and the agent now re-establishes its connection immediately when the gateway closes it, completing recovery in seconds rather than minutes. The agent's health status remains healthy throughout the reconnect, so the spurious alarm no longer fires. |
916659 | Fixed an issue where after an appliance upgrade, the Diagnostic Agent's manually disabled state was not preserved, causing unexpected error logs and alarms. Administrators who had intentionally disabled the service needed to re-disable it after every upgrade. |
848972 | When TSS is enabled and Large File Scanning (LFS) is enabled in the Netskope UI, but LFS is disabled in the DPoP CLI, the proxy still inspected files larger than 16 MB. This issue is now fixed. |
Here is the list of the new features and enhancements.
Increased SWAP Size
Increased the SWAP space to align with general recommendations and minimize memory performance pressure on the system.
Policy Support for HTTP Redirect
Netskope now supports a native HTTP Redirect policy action. This allows administrators to automatically route users to a new URL (e.g., a corporate app instance) using standard HTTP 307 headers, replacing the need for manually doing so via user notification links. To learn more, view Policy Support for HTTP Redirect.
Here is the list of fixed issues in this release.
| Issue | Description |
|---|---|
1031565 | After upgrading to appliance version 135.0.0, a timing issue during appliance startup could cause the URL database updater service to fail, resulting in the data plane (DPOP) stopping traffic processing. This occurred when an internal directory was created with incorrect permissions due to the order in which system components started up. This issue is now fixed. The required directories are now created with the correct permissions during initialization, ensuring the URL database updater starts reliably regardless of component startup order. |
1030520 | Fixed an issue where certain sites became unreachable due to an LFS configuration mismatch when LFS was enabled on the global tenant but disabled on the appliance. |
994372 | A high-severity "Diagnostic Agent Health" alarm appeared on the tenant dashboard roughly every 6 hours and cleared on its own after about 5 minutes. The diagnostic agent was healthy throughout β no data was lost and no functionality was affected. The alarm was a false positive. This issue is now fixed and the agent now re-establishes its connection immediately when the gateway closes it, completing recovery in seconds rather than minutes. The agent's health status remains healthy throughout the reconnect, so the spurious alarm no longer fires. |
916659 | Fixed an issue where after an appliance upgrade, the Diagnostic Agent's manually disabled state was not preserved, causing unexpected error logs and alarms. Administrators who had intentionally disabled the service needed to re-disable it after every upgrade. |
848972 | When TSS is enabled and Large File Scanning (LFS) is enabled in the Netskope UI, but LFS is disabled in the DPoP CLI, the proxy still inspected files larger than 16 MB. This issue is now fixed. |
Here is the list of the new features and enhancements.
Increased SWAP Size
Increased the SWAP space to align with general recommendations and minimize memory performance pressure on the system.
Policy Support for HTTP Redirect
Netskope now supports a native HTTP Redirect policy action. This allows administrators to automatically route users to a new URL (e.g., a corporate app instance) using standard HTTP 307 headers, replacing the need for manually doing so via user notification links. To learn more, view Policy Support for HTTP Redirect.
Here is the list of fixed issues in this release.
| Issue | Description |
|---|---|
1031565 | After upgrading to appliance version 135.0.0, a timing issue during appliance startup could cause the URL database updater service to fail, resulting in the data plane (DPOP) stopping traffic processing. This occurred when an internal directory was created with incorrect permissions due to the order in which system components started up. This issue is now fixed. The required directories are now created with the correct permissions during initialization, ensuring the URL database updater starts reliably regardless of component startup order. |
1030520 | Fixed an issue where certain sites became unreachable due to an LFS configuration mismatch when LFS was enabled on the global tenant but disabled on the appliance. |
994372 | A high-severity "Diagnostic Agent Health" alarm appeared on the tenant dashboard roughly every 6 hours and cleared on its own after about 5 minutes. The diagnostic agent was healthy throughout β no data was lost and no functionality was affected. The alarm was a false positive. This issue is now fixed and the agent now re-establishes its connection immediately when the gateway closes it, completing recovery in seconds rather than minutes. The agent's health status remains healthy throughout the reconnect, so the spurious alarm no longer fires. |
916659 | Fixed an issue where after an appliance upgrade, the Diagnostic Agent's manually disabled state was not preserved, causing unexpected error logs and alarms. Administrators who had intentionally disabled the service needed to re-disable it after every upgrade. |
848972 | When TSS is enabled and Large File Scanning (LFS) is enabled in the Netskope UI, but LFS is disabled in the DPoP CLI, the proxy still inspected files larger than 16 MB. This issue is now fixed. |
AISecOps DLP Agent enhancements
DLP Investigations & Quota Management
- New Tenant Default: Auto-investigation is now enabled by default for new tenants.
- UX Improvement: The investigation trail now allows for independent per-step expand/collapse, making it easier for analysts to focus on specific details.
- Attribution: Cases now surface βInvestigated byβ attribution, showing the user email for manual runs or βAutomatic investigationβ for system-initiated runs.
Case Details, Performance, and Context
- Enhanced Incident Context: Incidents originating from web browsing now display the Site name and URL in the incident side panel, giving investigators destination context.
- Speed & Accuracy: Significant performance improvements have been implemented for large tenants, including faster server-side duplicate-case search and a speedup on the Cases-list page count.
Integrations & Configuration
- Data View Schedule: Insight Generation cron is rescheduled from four times daily to daily at 2:30am PST.
- Rule Clarity: User-created Case Rules now show the authorβs email on the rule card, replacing the generic βCreated by Netskopeβ label.
- Safety Net: The rule editor now prompts the user to confirm discarding changes when canceling a form.
Documentation Link: The Help link in the agent UI now points to the public AISecOps DLP Agent documentation.
AISecOps DLP Agent enhancements
DLP Investigations & Quota Management
- New Tenant Default: Auto-investigation is now enabled by default for new tenants.
- UX Improvement: The investigation trail now allows for independent per-step expand/collapse, making it easier for analysts to focus on specific details.
- Attribution: Cases now surface βInvestigated byβ attribution, showing the user email for manual runs or βAutomatic investigationβ for system-initiated runs.
Case Details, Performance, and Context
- Enhanced Incident Context: Incidents originating from web browsing now display the Site name and URL in the incident side panel, giving investigators destination context.
- Speed & Accuracy: Significant performance improvements have been implemented for large tenants, including faster server-side duplicate-case search and a speedup on the Cases-list page count.
Integrations & Configuration
- Data View Schedule: Insight Generation cron is rescheduled from four times daily to daily at 2:30am PST.
- Rule Clarity: User-created Case Rules now show the authorβs email on the rule card, replacing the generic βCreated by Netskopeβ label.
- Safety Net: The rule editor now prompts the user to confirm discarding changes when canceling a form.
Documentation Link: The Help link in the agent UI now points to the public AISecOps DLP Agent documentation.
Content Redaction for AI Gateway
Sensitive data stays out of AI β automatically
AI adoption in enterprises is accelerating, along with a growing issue: employees paste sensitive data (PII, financial records, credentials) directly into AI prompts. AI Gateway 1.6 addresses this with Content Redaction: Data Leak Protection (DLP) masking intercepts sensitive content before it reaches the AI model and again on the way out.
Your DLP policies now extend all the way into AI prompts and responses.
No new tools. No new policy engine. If you already have Netskope DLP, it now works on AI traffic β automatically.

What you can do with it:
| Mask prompts | DLP policies detect and mask PII, credentials, and financial data in plain-text and JSON AI requests before they reach the model. |
| Mask responses | AI-generated responses are inspected and masked before delivery to the end user. |
| Plain text & JSON | Redaction works across plain-text prompts and structured JSON payloads. |
| Redaction event fields | New event fields show redaction status per transaction for audit visibility. |
For more information, see Redact Sensitive Data.
UX & Configuration Boosts
Streaming Response Inspection
Your security checks no longer come at the cost of user experience
Previously, AI Gateway buffered the entire AI response before running content inspection, so users saw nothing until processing finished. For long responses like code generation or summaries, this caused noticeable delay. AI Gateway 1.6 introduces streaming inspection: content is inspected and forwarded as it arrives, preserving the real-time feel of AI interactions.

AI Gateway-CLI Diagnostics
Run appliance checks from the menu β no shell access needed
AI Gateway 1.6 adds a Diagnostics section to the AI Gateway β CLI interactive menu. Administrators can select and run diagnostic commands directly from the TUI β without unlocking the appliance or accessing the underlying shell. Every run is recorded in the appliance audit log.
| Command | What It Tests |
|---|---|
ping | Network reachability β tests connectivity to a target host |
traceroute | Network path to a target host |
nslookup | DNS resolution and hostname lookup |
nc | Port connectivity β tests TCP/UDP connection to a host:port |
date | System clock and time-zone correctness |
Expanded AI Schemas & Providers
Three New Predefined AI Providers
More AI providers, same governance β no custom config needed
AI Gateway 1.6 adds three new predefined providers, giving your teams governed access to a broader set of AI platforms β without needing to configure custom providers manually. Select, apply policy, and go.

Custom Topics for AI Guardrails
Your guardrails, your rules β define content categories specific to your organisation.
AI Gateway 1.6 adds Custom Topics to AI guardrail profiles, letting you define organization-specific content categories for detection and policy enforcement.
Predefined categories donβt cover every organizationβs risks. Custom Topics let you tailor content categories to your industry, policies, or risk areas without waiting for Netskope.
For more information, see AI Guardrails Policy.
AWS Bedrock Mantle Endpoint Support
Access more models via AWS Bedrock β AI Gateway now supports the Mantle endpoint
AWS Bedrock exposes two distinct endpoint types with different host URL formats. AI Gateway v1.6 adds full support for the bedrock-mantle endpoint, which previously could not be used with Claude models.
| Endpoint Type | Host URL Pattern | Underlying Model | AI Schema | Example Provider |
|---|---|---|---|---|
bedrock-runtime | bedrock-runtime.{region}.amazonaws.com | Any | AWS Bedrock | cust-bedrock |
bedrock-mantle | bedrock-mantle.{region}.api.aws | Anthropic (Claude) | Claude | cust-mantle-cla |
bedrock-mantle | bedrock-mantle.{region}.api.aws | OpenAI-compatible | OpenAI (Compatible) | cust-mantle-ope |
For more information, see Configure AWS Bedrock as a Custom AI Provider.
Behind-the-Scenes Improvements
Custom Provider Hostnames Are Auto-Normalized
Envoy required lowercase. Now AI Gateway handles that for you.
Envoy, the underlying proxy in AI Gateway, requires strictly lowercase hostnames to route traffic correctly. Previously, configuring a custom provider with a mixed-case hostname could cause Envoy to fail routing silently β no error, just broken traffic. AI Gateway 1.6 automatically normalizes custom provider hostnames to lowercase before passing them to Envoy.

DLP Incidents Now Have Unique Titles
Goodbye, three identical incident titles
Incident objects now use the format, making each incident uniquely traceable. Previously, when multiple prompts were blocked by DLP in the same session, all incidents shared the same object name and title, making it impossible to match incidents to prompts.

ACTION REQUIRED
Enable Content Redaction
Content Redaction requires the DLP_ENABLE_REDACTION feature flag and must be activated.
- Confirm the feature flag is active:
GET /api/v2/policy/aig/dlp/featureflags. - Go to Policies β AI Gateway β DLP in your Netskope tenant.
- Ensure your DLP profiles include the data identifiers to redact in AI traffic (PII, financial data, credentials).
- Apply the policy to the relevant AI Gateway steering configuration.
Review AI Gateway-CLI Diagnostics Access
The new Diagnostics section in AI Gateway β CLI is available to all admins without unlocking the appliance. Review CLI access before upgrading if you enforce strict least privilege policies.
- Audit current AI Gateway β CLI admin access roles in your tenant.
- Confirm Diagnostics access aligns with your support and operations team scope.
OAuth Gateway URL Hardening
OAuth redirect URLs are now anchored server-side β attacker input ignored
The OAuth gateway URL is now anchored server-side; the HTTP/2 :authority header is ignored when building redirect URLs in the traffic intercept service. A security review found that an attacker could craft a malicious :authority value to hijack the OAuth flow to an attacker-controlled destination.

Mistral File Inspection Now Works Correctly
AI Gateway was speaking OpenAI. Mistral uses a different dialect.
The file ID detection logic now supports both OpenAI (file- prefix) and Mistral (UUID format) file reference formats. File content inspection for Mistralβs Files API failed silently because AI Gateway checked only for the OpenAI prefix file-. Since Mistral uses UUID-format IDs without a prefix, its file references were skipped by the inspection pipeline.

Content Redaction for AI Gateway
Sensitive data stays out of AI β automatically
AI adoption in enterprises is accelerating, along with a growing issue: employees paste sensitive data (PII, financial records, credentials) directly into AI prompts. AI Gateway 1.6 addresses this with Content Redaction: Data Leak Protection (DLP) masking intercepts sensitive content before it reaches the AI model and again on the way out.
Your DLP policies now extend all the way into AI prompts and responses.
No new tools. No new policy engine. If you already have Netskope DLP, it now works on AI traffic β automatically.

What you can do with it:
| Mask prompts | DLP policies detect and mask PII, credentials, and financial data in plain-text and JSON AI requests before they reach the model. |
| Mask responses | AI-generated responses are inspected and masked before delivery to the end user. |
| Plain text & JSON | Redaction works across plain-text prompts and structured JSON payloads. |
| Redaction event fields | New event fields show redaction status per transaction for audit visibility. |
For more information, see Redact Sensitive Data.
UX & Configuration Boosts
Streaming Response Inspection
Your security checks no longer come at the cost of user experience
Previously, AI Gateway buffered the entire AI response before running content inspection, so users saw nothing until processing finished. For long responses like code generation or summaries, this caused noticeable delay. AI Gateway 1.6 introduces streaming inspection: content is inspected and forwarded as it arrives, preserving the real-time feel of AI interactions.

AI Gateway-CLI Diagnostics
Run appliance checks from the menu β no shell access needed
AI Gateway 1.6 adds a Diagnostics section to the AI Gateway β CLI interactive menu. Administrators can select and run diagnostic commands directly from the TUI β without unlocking the appliance or accessing the underlying shell. Every run is recorded in the appliance audit log.
| Command | What It Tests |
|---|---|
ping | Network reachability β tests connectivity to a target host |
traceroute | Network path to a target host |
nslookup | DNS resolution and hostname lookup |
nc | Port connectivity β tests TCP/UDP connection to a host:port |
date | System clock and time-zone correctness |
Expanded AI Schemas & Providers
Three New Predefined AI Providers
More AI providers, same governance β no custom config needed
AI Gateway 1.6 adds three new predefined providers, giving your teams governed access to a broader set of AI platforms β without needing to configure custom providers manually. Select, apply policy, and go.

Custom Topics for AI Guardrails
Your guardrails, your rules β define content categories specific to your organisation.
AI Gateway 1.6 adds Custom Topics to AI guardrail profiles, letting you define organization-specific content categories for detection and policy enforcement.
Predefined categories donβt cover every organizationβs risks. Custom Topics let you tailor content categories to your industry, policies, or risk areas without waiting for Netskope.
For more information, see AI Guardrails Policy.
AWS Bedrock Mantle Endpoint Support
Access more models via AWS Bedrock β AI Gateway now supports the Mantle endpoint
AWS Bedrock exposes two distinct endpoint types with different host URL formats. AI Gateway v1.6 adds full support for the bedrock-mantle endpoint, which previously could not be used with Claude models.
| Endpoint Type | Host URL Pattern | Underlying Model | AI Schema | Example Provider |
|---|---|---|---|---|
bedrock-runtime | bedrock-runtime.{region}.amazonaws.com | Any | AWS Bedrock | cust-bedrock |
bedrock-mantle | bedrock-mantle.{region}.api.aws | Anthropic (Claude) | Claude | cust-mantle-cla |
bedrock-mantle | bedrock-mantle.{region}.api.aws | OpenAI-compatible | OpenAI (Compatible) | cust-mantle-ope |
For more information, see Configure AWS Bedrock as a Custom AI Provider.
Behind-the-Scenes Improvements
Custom Provider Hostnames Are Auto-Normalized
Envoy required lowercase. Now AI Gateway handles that for you.
Envoy, the underlying proxy in AI Gateway, requires strictly lowercase hostnames to route traffic correctly. Previously, configuring a custom provider with a mixed-case hostname could cause Envoy to fail routing silently β no error, just broken traffic. AI Gateway 1.6 automatically normalizes custom provider hostnames to lowercase before passing them to Envoy.

DLP Incidents Now Have Unique Titles
Goodbye, three identical incident titles
Incident objects now use the format, making each incident uniquely traceable. Previously, when multiple prompts were blocked by DLP in the same session, all incidents shared the same object name and title, making it impossible to match incidents to prompts.

ACTION REQUIRED
Enable Content Redaction
Content Redaction requires the DLP_ENABLE_REDACTION feature flag and must be activated.
- Confirm the feature flag is active:
GET /api/v2/policy/aig/dlp/featureflags. - Go to Policies β AI Gateway β DLP in your Netskope tenant.
- Ensure your DLP profiles include the data identifiers to redact in AI traffic (PII, financial data, credentials).
- Apply the policy to the relevant AI Gateway steering configuration.
Review AI Gateway-CLI Diagnostics Access
The new Diagnostics section in AI Gateway β CLI is available to all admins without unlocking the appliance. Review CLI access before upgrading if you enforce strict least privilege policies.
- Audit current AI Gateway β CLI admin access roles in your tenant.
- Confirm Diagnostics access aligns with your support and operations team scope.
Data access
Unlock more updates for Netskope
You're seeing 40 updates from the last 7 days without signing in.
Public session
Next step: Free - 7 days history and full feed tools. No card required to start.
What you can unlock
Higher tiers unlock more history and more rows.
Recommended next step
FreeSame 7 days window Β· adds plan perks
- Incidents, maintenance, updates, and advisories
- Unlimited vendor monitoring
- Analytics and trends
AlertsPlan perks
- Standard
30 days history
- Everything in Free
- Analytics
- Compliance snapshots
AlertsWider window
- Pro
90 days history
- Everything in Standard
- Analytics
- REST API
Wider window










