Main features of version 5.5

• Semantic AI engine: A new local Small Language Model (SLM)-based engine that classifies emails
  by meaning, not just structure. This enhances detection of sophisticated, well-written phishing
  or social engineering emails, even when traditional red flags are absent.
• Enhanced message scan indicators: message indicators now include both positive and negative
  signals and are integrated with the new "AI classifier engine" enabling more accurate and
  insightful threat assessments-making and response speed.
• API configuration for domain admins: Domain-level settings can now be managed directly via API,
  allowing greater automation and remote management for MSPs or multi-domain environments.
• AI dashboard widget: A new interactive widget displays real-time AI-driven threat classification
  activity, including drill-down views that help visualize detection trends and model classification
• Custom semantic categories: Each domain can now define up to eight context-aware threat categories.
  This allows admins to tailor the classification system to business-specific risks (e.g. invoice
  fraud) for sharper alerting and reporting.

Version 5.5.3 (Jun 23, 2025)

Improvements

• Dashboard: adaptive UI refresh, which reduces the HTTPs server loads and lets the session expire
• Dashboard: data ranges have more "last X hours" ranges and no longer allow week or month ranges
• Dashboard: show explicit warning when the license is expired
• Dashboard: greatly optimize the database performance of the "analyzed message", "email flow", "threat maps", "cluster status", "AI classification", "message distribution", "top sender" and "top recipient" widget
• Licensing: show previous license information when the license is expired
• Microsoft Azure: triggers a post-deployment storage optimization procedure after every storage expansion
• Microsoft Azure: upgrade agent version to 2.13
• Search: added "last 6 hours", "last 12 hours" and "last 48 hours" filters
• Web UI: temporary disable form on submission to avoid multiple actions caused by double-clicks

Bug fixes

• Email Continuity: fixed mail compose editor when logged in as user with role "basic user"
• License accounting: excludes released messages from accounted deliveries
• Message details: fixed spam confidence when spam reports contains multiple rules with the same name
• Syslog: fixed logging to syslog of some Web-UI failed login attempt

Version 5.5.2 (Jun 12, 2025)

Security

• Notification to sender: notifications are sent only for outgoing messages

Improvements

• Microsoft Azure: added post-deployment storage optimization procedure
• Microsoft 365: added “hybrid” label in test page for users that are hybrid and not fully migrated
• Microsoft 365: fixed import of user with duplicated additional addresses
• ESG AI: fine-tuned engine parallelism when load average is high

Bug fixes

• File type: updated blocked audio and video label descriptions
• Message details: show when failed DMARC results are ignored since a message is sent by trusted sender
• Microsoft Azure: fixed firewall rule for Microsoft management IP
• Outlook add-in: fixed release action when a reason is required
• Search: fixed export of saved search results
• User management: fixed user export for domain admin
• User management: permit to “reset to default” customized user defaults

Version 5.5.1 (May, 26 2025)

Security

• Whaling protection: ignore outgoing email from SPF validated Google Workspace tenants

Improvements

• Microsoft 365/Google Workspace: convert local users as functional, when a Google user is converted to group
• Microsoft 365/Google Workspace: when a local user is converted to functional (e.g, shared mailbox,
  groups), all emails assigned from previous group membership are removed

Bug fixes

• Google Workspace: fixed creation of functional users when group emails aren't assigned as user aliases
• Google Workspace: fixed an exception when searching by address
• Greylisting: avoid duplicated entries for auto welcomelisted senders
• HTTP: avoid server name canonicalization for redirect when using self-signed certificates
• Outlook addin: fixed report as bad action
• Quarantine digest: null users and null functional users roles are ignored when generating reports

API

• ADD: added /network/dns-forwarder to configure DNS forwarder
• ADD: added /network/static-host-lookup to configure Static Host Lookup
• ADD: added /sender-dependent-relay to configure sender dependent relay

Version 5.5.0 (May 12, 2025)

Security

• WAF: improved application firewall blocks for API access
• URLSand: improved "base URL" disarming
• URLSand: improve scheme parsing for HTTP/HTTPS

Improvements

• Mailer: added "%org-name%" header to all ESG generated emails and reports
• Mail scanner: fine tuned configuration to avoid cpu saturation
• Message details: show multiple SPF, DKIM and DMARC authentication results with extended information
• View mail: show CC header in the header section
• User management: user export will include email address and domain admin permissions
• User role: allow customization of "can view" capability for "clean" result category

Bug fixes

• Message details: show when welcomelist are ignored due to DKIM or DMARC failures
• Message details: don't store "cid:" links and http local anchors
• Search: apply custom permissions for message results to custom administrator roles
• System preferences: removed empty section "System notification" in cloud appliances
• Rebranding: replaced "Antispam Service" with "Email Security"
• Saved search: disambiguate duplicate names

API

• ADD: added /machine-learning/ai-classifier to configure AI Classification engine
• ADD: property aiClassifier to /message/{id}
• ADD: property hostname to /integration/pop3
• ADD: added /trusted-network to configure trusted networks
• ADD: property headers.spfValid to GET /message/{id}
• ADD: added headers.authenticationResults to GET /message/{id}
• IMPROVED /message/{id}/fetch returns 410/Gone when a message is removed or has been rotated
• IMPROVED: property headers.dkimValid in GET /message/{id} will honor alignment to sender domain
• IMPROVED: properties headers.spfResult/headers.spfIdentity from GET /message/{id}
• FIXED: removed POST/PATCH/PUT action on non-editable /quarantine/release-request-entry
• FIXED: removed /graph since it's included in /report output
• FIXED: allow /report to domain administrator and user roles
• FIXED: filter sender/recipient in GET /restricted-sender
• FIXED: filter sender/recipient in GET /attachment-filter/file-name-rule,
  GET /attachment-filter/file-type-rule, GET /attachment-filter/password-protected-archive-rule
  and GET /attachment-filter/scan-archive-rule
• FIXED: avoid duplicate entries with the same domain unique property for /relay
• DEPRECATED: properties headers.spfResult/headers.spfIdentity from GET /message/{id}

Breaking changes

There are no breaking changes in this release.

Version 5.4

All upgrades from previous versions are included. See the full release notes of Libraesva ESG version 5.4.