The Run Risk Assessment flow lets you launch a risk assessment against one or more AI systems or datasets using a predefined template. You can run as many subjects as you like in a single batch, and choose whether to complete each assessment yourself in the platform or send it to someone outside your team using a secure link. This helps ensure consistent evaluation of risks such as bias, privacy, quality, and more.
Step 1: Select a template
Choose a template from the list. Templates define the type and scope of the assessment (for example, Bias and Fairness, or Privacy Impact). Each template shows:
- What it applies to — AI systems, datasets, or both
- Frameworks it maps to (such as NIST AI RMF, EU AI Act, or ISO 42001)
- Estimated time to complete and the number of questions
The list is tailored to your environment, so you'll only see templates that are relevant to what you're assessing.
Once selected, click Next.
Step 2: Select subjects
Select the AI systems or datasets you want to assess. The available list matches the type supported by the template you chose in Step 1.
You can select multiple subjects at once. Any subject that has already been assessed with this template is hidden from the list, so you won't accidentally create a duplicate.
You can search or scroll to find the right subjects. Once selected, click Next.
Step 3: Choose how to run it
Decide how each assessment should be completed:
- Complete in the platform — assessments are assigned to you and appear in your Risk Assessments dashboard, ready to start.
- Send externally — send a secure link to someone outside the platform (for example, a vendor or a colleague in another team). For each recipient you can add a name and email address, an optional message, and a due date. Recipients open and complete the assessment through the link without needing a login.
Once you've chosen, click Next.
Step 4: Review and Create
This screen summarises what you're about to create:
- Template name
- Number and type of selected subjects
- Delivery method (in-platform or external link), and recipients for external assessments
- Total number of assessments to be created
Click Create to create the assessments. Everything you create together is grouped as a single batch so you can track it as one unit.
- In-platform assessments appear in your Risk Assessments dashboard, where they can be viewed, continued, or completed.
- External assessments are emailed to each recipient as a secure link. You can resend a link at any time from the dashboard.
Tracking a batch
Assessments launched together appear as a batch on the Risk Assessments page. Each batch shows the template, the total number of assessments, how many are done and in progress, when it was run, who ran it, and the delivery method. External batches are marked with a Sent to external label and link through to the matching external assessments.
Reviewing risk levels
When an assessment is completed, the dashboard shows two risk levels side by side:
- User assigned — the risk tier set on the AI system or dataset record
- System recommended — the risk level the completed assessment suggests, based on the answers given
Showing both together makes it easy to spot a mismatch — for example, where a system was recorded as Moderate risk but the assessment recommends High. You can reconcile the recorded risk tier to the recommendation in one click from the AI system's detail page.
Notes
- Only templates relevant to your subject type are shown during selection.
- Each subject you select creates a separate assessment using the chosen template.
- Subjects already assessed with the selected template are excluded automatically to prevent duplicates.
- You can return to an in-platform assessment later using the Continue action in the dashboard.
- For external assessments, use Remind in the External Assessment dashboard if a recipient needs the link again.
Related Help Pages
Automated Compliance (Risk Assessment Recommendation Engine)