Note: This article is updated weekly throughout the month as new SafeBase features are released. All updates are also published on our Product Updates site.
Table of Contents
New Graph on AI Analytics page: Answering Methods graph
A new Answering Methods graph is now available on the AI Analytics page, located under Dashboards in SafeBase. This page helps customers track how many questions are answered
Questionnaire Owner Field
You can now assign a Questionnaire Owner in SafeBase to clearly designate who’s responsible for managing a questionnaire.
How it works
Assign or reassign ownership to any existing SafeBase user when uploading or editing a questionnaire.
The assigned owner receives an email notification when they’re designated.
Available now in the SafeBase app (coming soon to the Chrome Extension and other intake points).
(BETA) Approve and Create Accounts in Teams
Previously, Microsoft Teams approvals only worked for existing accounts, requiring users to switch to the SafeBase app to complete requests for new accounts, an extra step that created friction. You can now create new user accounts when approving access requests directly in Microsoft Teams.
BETA INTEGRATION:
This integration is currently in BETA and requires enablement by the SafeBase team. To request access, please contact your Customer Success Manager
Note: All Teams customers automatically receive this update with any version of the SafeBase bot.
SafeBase Partner Settings for Drata VRM Access
SafeBase Partner settings now let administrators control how Drata VRM accesses Trust Center content. This ensures vendors stay in control while enabling upcoming automation from the Drata VRM Agent.
Solution Details
The SafeBase Partner settings provide this essential control:
Administrators can decide whether Drata users can request access to private Trust Center content directly from VRM or if they are limited to only publicly available content.
Access requests made via Drata will be logged in the Accounts page in SafeBase and will clearly display Drata VRM as the source.
Limitations
By default, Drata VRM Partner settings allow access to both Public and Private content, while still requiring all access conditions (e.g., NDAs, Terms of Service).
Initial Availability: Access requests will only be available to customers who are part of the Drata VRM Agent Closed Beta.
Microsoft Teams AI Search
Customers can now ask security questions directly in Microsoft Teams and get AI-generated answers from their Trust Library. This makes it easier to collaborate and respond quickly without leaving your workflow.
BETA INTEGRATION
This integration is currently in BETA and requires enablement by the SafeBase team. To request access, please contact your Customer Success Manager
Solution Details
In any message in Teams, customers can choose to ask a question to SafeBase and they will get an AI-generated answer using sources from their Trust Library
The answers will respect the role and permissions of the linked user in SafeBase.
The answers use the same mechanism as the Trust Library AI Search and Slack AI Search.
They can optionally choose to post the question/answer anywhere in Teams (such as a DM or channel).
Limitations
New bot upload required: Teams admins must upload a new version (1.0.2) of the bot to use this feature.
AI must be enabled: Generative AI must be turned on in Settings.
This BETA Integration is only open to paid plan customers only.
Trust Center: Enhanced AI Card & New AI Feature Item
We’ve expanded the AI card and added a new AI Feature item in the SafeBase Trust Center to help customers clearly explain how AI is used across their products and operations. These updates help Trust Center visitors find the AI answers they need. Learn more at Trust Center - AI Feature Items.
Expanded AI Card
We’ve added two new standard sections designed to provide meaningful context to Trust Center visitors:
AI Overview: Share a high-level summary of how AI is used within your organization, across products, operations, or internal workflows, and communicate your approach to responsible AI adoption.
Evaluation & Red-Teaming: Highlight your AI safety practices, including model evaluations, adversarial testing, jailbreak prevention, and other risk mitigation measures that are increasingly expected during security reviews.
AI Feature Item
You can now showcase individual AI-powered features or products directly within your Trust Center. Each AI Feature item lets you:
Explain what the feature does and how AI powers it.
Specify the underlying AI models in use.
Link to model documentation or performance benchmarks.
Attach diagrams or documents showing how data flows through the AI system.
Trust Center AI Search
We’re introducing Trust Center AI Search, a new way for visitors to quickly find the security information they need. Instead of relying on keywords, customers can now ask questions and receive either the most relevant results or a direct answer with the “Ask AI” mode. Learn more at Trust Center - AI Search.
What’s New
Natural-language questions: Ask in plain English, like “How do you encrypt customer data?”
Ask AI mode: Get clear responses based on the Trust Center content available to you (disabled by default and can be enabled in Settings > Generative AI)
Note:
Rollout begins today with Enterprise accounts and will expand more broadly over the next two weeks.
AI Search respects all existing Trust Center access controls. Visitors only receive results and answers based on the content they are authorized to view.
Automatic Name Redaction (KB)
SafeBase now automatically redacts customer names from questionnaire responses before they are added to the Knowledge Base. This prevents sensitive details from appearing in AI-generated answers and saves customers from manually reviewing and editing entries.
What’s New
Automatic redaction: Customer names are removed before responses are added to the Knowledge Base
Safe replacements: Names are replaced with generic terms to preserve response accuracy
AI-ready content: Keeps Knowledge Base entries clean for AIQA, Trust Library Search, and AI Search
Notes
Applies only to questionnaire responses being added to the Knowledge Base
Requires ‘Enable Generative AI’ to be turned on (already enabled if using AI Questionnaire Assistance)
Streamline Questionnaire Uploads and Portal Access
We’ve made small but helpful improvements to questionnaires in SafeBase. These changes make it clearer when spreadsheets are uploaded and reduce confusion when working with portal links.
What’s New
Auto-detect notifications: Customers are now notified when spreadsheet columns are automatically mapped during upload
Clearer portal button: The “Start” button for portal URLs now says “Open” with a tooltip guiding customers to use the Chrome Extension
CRM Sync Accuracy with Custom Tiebreakers
We’ve expanded CRM integrations to support custom tiebreaker rules during hourly account syncs. This helps ensure the right CRM account is linked in SafeBase, improving automation and reporting accuracy.
What’s New
Custom tiebreaker rules: Apply logic such as “most recent Last Modified Date” to resolve duplicate CRM accounts
Automatic linking: Hourly syncs now use these rules to connect SafeBase accounts with the correct CRM accounts
Notes
Available only on SafeBase Advanced and Enterprise plans
Custom tiebreaker rules must be configured by the SafeBase Solutions Architecture team