Data & More

Release Note v26.5.0

7 min readJul 15, 2026

Aggregated changes across patches v26.4.1 through v26.4.24: new DLP enforcement module, LLM document classification, AI dashboard charts via MCP, mailstore (IMAP) source type, data-subject tagging, and platform hardening.

Release date: 1 Augus 2026| Previous version: v26.4.0 (15 May 2026)

Scope: Aggregated changes across patches v26.4.1 through v26.4.24 and the upcoming v26.5.0 minor cut.


Features

DLP enforcement module (new service)

  • New standalone DLP service with a dedicated DLP page in the client (Settings, Transport Rules and Audit tabs)

  • Exchange transport-rule enforcement managed from a central DLP configuration, with one-click automatic enabling of transport rules (MS OAuth handler plus client button)

  • Per-algorithm-category transport-rule switches (new dlp_transport_rule flag on algorithms and algorithm_category); incompatible algorithms are blacklisted automatically

  • Document classes can now trigger DLP actions via new dlp_notify and dlp_block flags

  • Teams notifications: dedicated Teams bot with automatic installation, notifications for all sources, and a direct link to the Teams chat message in each notification

  • Enforcement flow: grace periods before enforcement, admin enforcement and cancellation, revert-block action with grace-period options, and an exclusions table with removal support

  • Graph integration: OCR is invoked for DLP processing; files not shared externally or last modified by an application are skipped; SharePoint "all" scope excludes OneDrives; quarantine email returns the moved message id

  • Full audit trail in the new dlp_audit index, including sender name and email; redesigned DLP audit tab with chart, summary cards and polished log table

  • New "Manage DLP" IAM permission; all DLP routes are permission-protected

LLM document classification

  • LLM classifier integration in document-class tagging (new llm-management service integration)

  • LLM classification controls in the Document Class Builder, including a new output_class_only document class type

  • All marking and refresh tasks respect the llm_class flag; classification calls authenticate via IAM_SECRET

AI dashboard charts and MCP server

  • Built-in per-tenant MCP server for AI-created dashboard charts

  • MCP key management in the client with a new "Manage MCP Keys" IAM permission

  • AI charts render on the dashboard (new ai_chart_ids field on dashboard configs); new analytics routes serve AI chart data

Mailstore (IMAP) source type

  • New mailstore source type backed by the IMAP service: creation form, settings page and account management

  • Accounts are set active on registration; account list with metadata-only enforcement indicator

  • Mailstore enforcement routing through the IMAP service

Data-subject tagging (Known Persons)

  • Tag data subjects directly: new tags on the known-persons index with status per tag

  • Quick-create a tag from a data subject, with custom tag naming

Tagging engine: label-value matching

  • New Label Value matching in tags, including multiple label values per tag and "label value near label value" queries

  • Near-text tag queries now support multiple words

  • Runtime-mapping efficiency improvements for NearLabel and LabelValueNear

  • New tagger handler for the label_value type in the profiler

Token and drive fishing (Graph pre-filtering)

  • New FISHING ingestion mode for token-based pre-filtering of Graph sources, with KQL query batching and time bounding

  • Drive fishing ingestion service with diff-based new/delete detection and false-positive token exclusions

  • Enabled per source definition; fishing queries are loaded from a storage file

MIP label hierarchy

  • Parent information is stored for MIP labels; the client shows the full label hierarchy

  • Parent labels are hidden in policy enforcement selection

  • Both the new and the old MIP label format are written when tagging

  • MIP tagging is disabled automatically when enforcement is disabled on the source

Google Drive and Gmail in Data Owners

  • Google Drive and Gmail sources now appear in Data Owners

  • Owners and custom owners on Drive sources (new owners field on google_drives); owners shown instead of creator id

  • Custodian spreadsheet download and upload for Google Drive sources

OCR pipeline

  • Combined worker for signature, YuNet and ID-or-not processing

  • New tesseract-only extraction route with a dedicated express queue

  • Direct ingestion respected end-to-end; PDF images with annotations over them are skipped; unreadable PDF images no longer crash processing

Client platform overhaul

  • New DropdownPanel component adopted across the entire app: teleported panels, smart positioning, unified styling

  • Bootstrap fully removed from the client (drained into scoped SCSS partials, modern CSS reset, stylelint added)

  • Logs moved to a separate sidebar page

  • TinyMCE editor removed; new HTML/Markdown conversion utilities


Changes

Credentials and app registrations

  • Daily reminder emails to sendout managers for 14 days before an Azure app-registration secret expires

  • Secret-expiration checkbox on the app registration form

Purge and reingestion

  • Purge-all support for Outlook and OneDrive graph_modern sources; purge status criterion with migration

  • Stale graph_modern SharePoint sites are tagged for purge; purge cleaning alert added

  • Case-insensitive site and drive matching in purge; obsolete reactive purge code removed

Source defaults

  • Graph sources: default deltas set to daily, default threads reduced from 8 to 2

  • Fileshare sources: default one thread per source

  • New sources get updated defaults for shared and personal mailbox ingestion

Data Owners

  • Multiple emails per path via semicolon-separated values in custom owners and Outlook mailbox uploads

  • Additional columns supported in custom-owners uploads; upload lines sorted before parsing

  • Custodian uploads standardised on spreadsheets (CSV upload disallowed in custodians)

  • "Show more" label reflects 10,000+ results

Exchange on-premise (EWS)

  • Delete-tagged enforcer with skip-orphan support

  • Business-hours throttling driven from the collector

  • Duplicate-account guard on account creation

Reporting and analytics

  • ITR variant of the PoC report, plus a classified-documents-per-language section

  • PoC and monthly report colours aligned with the dashboard; key figures hidden when SharePoint data is missing

  • Executive summary reworked; groundwork for AD-group scoped monthly reports

  • Usage stats switched to percentages; server-stats history kept for 14 days

Platform resilience

  • Elasticsearch bulk writes no longer trigger 429 storms

  • Graph 429/503 responses retried honouring Retry-After

  • ES health alerts tuned: red-only on single-node clusters, red-or-yellow on multi-node clusters

  • Tika streams auto-closed to prevent leaks (EWS, IMAP, Google, fileshare, Graph)

Integrations

  • Simplified DMCS to Redaction Tool integration (push model); Pixel Redaction button in the REDACTED tab

  • Deploy GitHub Action with Ansible playbook for orchestrated host upgrades; release action can be triggered by the management tool


Bugfixes

API

  • Fix race conditions when marking fileshare sources and in CSV creator/owner update tasks

  • Fix sort order when fetching Data Owners documents; fix sorting report documents by path

  • Fix user-policies aggregation and report-sendout blacklist edge cases

  • Fix Azure user search count mismatch; apply search query to groups field; group searches made case-insensitive

  • Fix ES version issue for counts with runtime mappings

  • Fix LLM classification scope and stale-detection in unmark tasks

  • Fix missing HubSpot id in caselog contacts; fix edit-exchange for demo accounts

Client

  • Post-migration dropdown fixes: overflow, stale search, double borders, hidden panels inside teleported containers

  • Fix empty groups table in user settings; fix login language switcher

  • Fix flicker in the notification report; fix multiselect input position

  • Fix report file download permissions; fix copying report document paths for Gmail sources

  • Fix app-registration error when no groups are selected

  • Fix orphans-only filter for nested folders in Data Owners

Graph services

  • Fix stale owners on SharePoint validate; prefer the SharePoint owners group for site owners; derive site owners from the full root-item ACL

  • Fix cross-tenant site deletion during full ingest

  • Fix drive-fishing filter dropping search results

  • Fix HTML-response crash in mailbox settings fetch; fix OkHttp connection leak in enforcer file download

  • Fix DISC to DISP typo in the move enforcer (EWS, IMAP and Graph)

  • Return 404 instead of an error when a user photo is missing

IMAP / Mailstore

  • Fix environment-variable loading for the Elastic config

  • Set in-ingestion status on new attachments; log authentication issues

Analytics

  • Fix division-by-zero and missing-data edge cases in monthly report generation

  • Fix external data queries and queries that did not specify an index


Security hardening

  • Rate limiting: NGINX rate limiting added on authentication routes across all stacks

  • Permission model: new "Manage DLP" and "Manage MCP Keys" IAM permissions; DLP routes permission-protected; logs permissions moved to a dedicated category; header links restricted by organization permission

  • Service-to-service auth: internal /classify calls authenticate with IAM_SECRET in the Authorization header

  • Audit trail: DLP audit logs record actor name and email; audit-log sender fields reworked

  • XSS surface reduction: TinyMCE editor removed from the client; HTML sanitisation consolidated

  • Tenant isolation: cross-tenant site deletion during full ingest fixed

  • Credential hygiene: proactive secret-expiration reminders for Azure app registrations; last-login schema strips invalid emails


Was this article helpful?