WAI Docs Wed Aug 19 13:22:37 EDT 2026
List
Quick Start
Quick Start
User Guide
User Guide
Policies - GuardRails
Policies - GuardRails
Witness Anywhere: Remote Device Security
Witness Anywhere: Remote Device Security
Witness Attack
Witness Attack
Administrator Guide
Administrator Guide
404
404
WitnessAI Release: July 21, 2026
Note: New and updated features may not be immediately available on your dedicated deployment due to participation in limited availability releases, beta programs, or regional configurations.
If you don't see a feature you expect, please reach out to Customer Success. We're happy to help!
New Features
Salesforce AI Agents
- Salesforce AI Agents is now supported for Prompt, Response, & Tool Action observability. Policies support “Salesforce AI” as an Application Destination.
Figma AI (Design Agent)
- Figma AI, including the Design agent, is now supported for Prompts and Responses. Policies support “Figma AI” as an Application Destination.
Feature Updates
Expanded Routing Support
- Routing is now supported as a Guardrail Action for the following applications: Microsoft Copilot (Corporate and Personal), Microsoft Office, Codex desktop app, Codex web (chatgpt.com/codex), and VS Code when using Codex models.
Zscaler OneAPI Integration
- The firewall integration now supports Zscaler's OneAPI for pushing URL category and FQDN object updates. Organizations using Zscaler can use this API path in addition to the existing integration methods.
Performance, Reliability, & API Updates
No Performance, Reliability, & API Updates in this release.
Fixed Issues
Gemini: DLP policy was not reliably blocking attachments uploaded in Gemini sessions. Attachments are now correctly evaluated and blocked when policy requires it.
MCP Tool Actions: A recent change to Claude's MCP naming convention caused tool-name policy matching to fail to match to legacy MCP Server name formats. stop working. Accurately matching MCP tool calls to MCP Servers is now restored.
Attachments: Data protection policy was not being applied to .txt file attachments. Plain text files are now correctly inspected and enforced.
Attachment Scanning: The pre-prompt safety filter was running on messages even when the policy inspect_attachments setting was disabled. Disabling attachment scanning now works correctly.
Grok Warning Messages: were not reliably displayed in the Grok user interface. Warning messages now display correctly.
Known Issues
No Known Issues in this release.
Breaking Changes
No Breaking Changes in this release.
Please contact Customer Success if you have questions or need support with this release.