Update AIG-CLI wording

The AI Guardrails service is formally renamed to Configure Guardrails LLM to more accurately represent its functionality and intended use.

Netskope has extended the functionality of three attributes to support authorization to the Guardrails deployed on GCP TPU.

Client ID: Provide an OAuth2 client ID if using an external provider, or a service account email if deploying to a TPU; leave blank if not required.
Client Secret: Enter your OAuth2 provider secret or, if deploying to a GCP TPU, the associated RSA private key.
Enter new scope: Specify the scope for standard token requests or the JWT audience required for TPU deployments.

Introducing GCP to AIG version 1.3.338 (Beta Version)

Netskope is excited to announce extended infrastructure support for Google Cloud Platform (GCP). Customers can now deploy Netskope virtual appliances directly within their GCP environment using pre-configured VM images. This update provides GCP-centric organizations with the same seamless traffic steering and data integration capabilities previously available for AWS.

Custom Rate Limit Response Messages

Netskope has added greater flexibility by allowing you to define custom response messages for rate-limited policy.

Previously, the system triggered a hard-coded response whenever a user exceeded their quota. Now, you can inject specific instructions, brand-aligned messaging, or direct support links into the response body. This shift empowers your developers to self-solve issues and reduces support overhead.

Improved navigation – On boarding page is now available under AI Gateway section

Netskope has moved the AI Gateway registration page from Settings > Security Cloud Platform > On Premises > On-Premises Infrastructure > Next-Gen to Settings > Security Cloud Platform > AI Gateway > VM Onboarding.

Extended Support for AI Schemas

Netskope has expanded AI Gateway to provide basic functionality for unsupported AI schemas beyond OpenAI, Claude, and Gemini. It now allows you configure, authentication, rate limiting, and access control for all schemas.

This feature does not include content inspection for the unsupported schemas.

Netskope has streamlined the UI for providers using the other schema to ensure configuration accuracy. These providers are now hidden from DLP and AI Security policies, and the Model field is automatically removed for Access and Rate Limit policies. In addition, the interface now explicitly clarifies that only the other activity type is supported for these providers, preventing unsupported configuration attempts.

Added new predefined AI providers with OpenAI compatible schema

Netskope now includes predefined AI providers with OpenAI compatible schema. You can configure and connect your OpenAI SDK with those predefined providers with the below endpoints directly.

xai-openai: xAI compatible with OpenAI

Limitation: AI Gateway blocks chat completions requests with the deferred parameter because it deviates from the OpenAI spec and causes unexpected responses.
deepseek-openai: DeepSeek compatible with OpenAI.
mistral-openai: Mistral AI compatible with OpenAI.

Limitation: Content inspection does not support file downloads (GET /files/{id}/content ) due to an unexpected file_id format.
together-openai: Together AI compatible with OpenAI.

Netskope only supports API sets that are compatible with the Open AI schema.

AI Gateway Release Notes Version - 1.3

Release Notes Subscription