Re-evaluating assessments

Privacy assessments reflect the state of your systems at the time they were generated. When system data changes (new data categories are discovered, data use declarations are updated, or datasets are modified), the AI-generated answers may no longer be accurate. Re-evaluation lets you refresh only the affected answers without discarding your human edits.

When to re-evaluate

Re-evaluate an assessment when:

• A system's data categories have changed (e.g., a new scanner run discovered additional categories).
• Data use declarations have been added, removed, or modified.
• The system's processing purposes or legal bases have changed.
• The assessment status shows Outdated.

Fides marks an assessment Outdated automatically when it detects that the system snapshot stored with the assessment no longer matches the current system state. You'll see the Outdated badge in the assessment header whenever this is the case.

How change detection works

When you open an assessment (or when Fides performs a background check), it compares:

• The current system metadata against the snapshot stored when the assessment was generated
• Current data use declarations against the stored declarations
• Current data categories against the stored categories

If any of these differ materially, the assessment status changes to Outdated. Fides tracks which question groups are affected by the changed fields so that re-generation only targets the relevant questions.

Running re-evaluation

To re-evaluate assessments, select the outdated assessments from the assessments list and click Re-evaluate. Fides queues a background job for each assessment and processes them in parallel.

Re-evaluation:

1. Reads the current system data and computes a new privacy context.
2. Identifies which questions are affected by the changed data.
3. Re-runs the AI only for affected questions.
4. Creates new answer versions with source "Agent" for any regenerated answers.
5. Updates the assessment snapshot to reflect the current system state.
6. Changes the assessment status back to "In progress".

Human edits are preserved

Re-evaluation never overwrites answers that a human has edited. If you edited an AI-generated answer and the underlying data later changes, re-evaluation creates a new AI-generated version in the history but your edited version remains the current answer. You can review the new AI suggestion in the history view and decide whether to adopt it.

Automatic reassessment

You can configure Fides to run re-evaluation on a schedule using the automatic reassessment setting. When enabled, Fides periodically compares all active assessments against the current system state and queues re-evaluation tasks for any that are outdated. See configuration for how to enable and schedule automatic reassessment.