September 29, 2026: WitnessAI Release

WitnessAI now supports prompts, Intention, and responses for Databricks Genie Deep Research.

The User, Group, and Application filters on the Alerts and Conversations pages now support selecting multiple values at once, so you can view combined results for several users, groups, or AI apps in a single view. Filters include Select All, Clear All, and Invert actions, and filter settings can be saved as a named filter set for reuse.

A new Behavior Type column shows the specific Behavioral Activity or Model Identity Protection rule that triggered each alert. You no longer need to open each alert to see which rule fired, which makes reviewing alert accuracy much faster.

When there are more filters than fit on one row, the filter bar now wraps instead of cutting filters off. You can use the new customize option to pin, unpin, and reorder filters, and your layout is saved for your next visit. Severity, User, Group, Application, Guardrail, Risk Type, Action, and Policy are pinned by default. Data Type, Label, and Capture Time are available from the customize menu.

Every chart on the Spend Overview and Metered Usage views now has an Export CSV option, so you can share or analyze the data outside the console. The Tokens by User chart exports the top 100 users by token volume. All other charts export their full data series.

The Value dashboard header now matches the layout of the other Home tabs. A Clear button appears when any filter is active, and the filter controls wrap more cleanly on smaller screens.

No Performance, Reliability & API Updates in this release.

Fixed an issue where a policy could be saved while an enabled guardrail on another tab still had unresolved validation errors.

Fixed an issue where updating a Behavioral Activity guardrail's purpose or rules could in some cases clear the configuration of other guardrails in the same policy.

Fixed an issue where filtering the user list by role could use a user's assigned role instead of their effective role, which could result in an incomplete list for Admin and SuperAdmin users.

Fixed an issue where sensitive data on the Users tab could show a placeholder instead of the masked value, and the reveal (eye) icon might not display the data.

Attachments uploaded to ChatGPT may be allowed even when a policy is set to block them.

Microsoft now redirects copilot.microsoft.com to copilot.com. Prompts sent from copilot.com can show as Microsoft Copilot (Corporate) in the console instead of Microsoft Copilot (Personal), and policies scoped to Copilot Personal might not be applied to them.

Prompts sent from the ChatGPT add-in in the Excel desktop app may not be captured, so they might not appear on the Conversations page and policies might not be applied to them.

Prompts sent from Notion AI may not be captured. When this happens, the prompts might not appear on the Conversations page and policies might not be applied to them.

No Breaking Changes in this release.

Please contact Customer Success if you have questions or need support with this release.