v2.3.1 - May 5th
2 min
this release enhances moderation workflows with a dedicated appeal console featuring a familiar interface, introduces rule testing to validate moderation policies before deployment, and includes important s ecurity hardening and reliability improvements new features appeal moderation console a new dedicated console for handling appeals now provides a ui similar to the reports moderation console, making it immediately familiar to your team all appeals from reporters or reported accounts are consolidated into a single view , and previous moderation actions along with any decisions following appeal are displayed with greater clarity this reduces context switching and makes it easier for moderators to understand the full history and context of each appeal rule testing you can now test moderation rules with individual content before deploying them at scale from actions settings test one piece of content at a time to validate rule logic, prioritisation, and routing to queues and see clearly confidence scores assigned by ai providers as well as how nima would ingest a similar piece of content this prevents unintended consequences of policy changes and ensures rules work correctly before affecting live moderation workflows improvements strike system hierarchy in actions page when viewing moderation actions on the actions page, you can now see clear breadcrumbs displaying the strike tier hierarchy this gives moderators and reviewers immediate context about the enforcement level and helps ensure proportionate action application default granular quality control permissions new environments now come with granular quality control permissions pre configured, eliminating the need for manual setup quality control teams can immediately begin monitoring moderation decisions without administrative delays enhanced feedback messaging improved content safety policy feedback messaging to provide clearer guidance to moderators during decision making bug fixes & security updates user management fixed a client side exception that occurred when editing users with no assigned role, preventing configuration errors during team setup ai detection integration corrected sightengine ai model output handling to properly display classification scores instead of defaulting all offensive category scores to 1 user notifications prevented automated detection reporters from receiving email notifications, reducing unnecessary notification volume