Set up your organizationConfigure masking
Configure masking
Protect sensitive fields in your datasets by anonymizing, pseudonymizing, or hiding them.
With Masking you control how Power BI Agent handles sensitive data, on a per-field basis. This is available on Plus and Premium plans.
How it works
Masking operates at the column level within your dataset. For each column, you choose one of three options:
| Option | What happens |
|---|---|
| Anonymize | The value is replaced with '***'. The field can no longer be read or used by the AI. |
| Pseudonymize | The value is replaced with a consistent pseudonym. The AI sees the pseudonym, so the field can still be used for analysis. In the chat you see the real values yourself. |
| Hide table | The entire table is not shown to the AI. Use this for tables that are not relevant for analyses. |
Masking is applied before results are sent to the AI. The original data in Power BI remains unchanged.
Configure masking
- Go to Organization > Datasets.
- Click the menu on a dataset and choose Masking.
- For each column, select the desired option: Anonymize, Pseudonymize, or Hide.
- To hide an entire table, select Hide table at the table name.
- Click Save.
When to use which option
- Anonymize: personal data, email addresses, social security numbers — fields you never want to appear in analyses.
- Pseudonymize: products, company names, or relation numbers where you do want to analyze on the field, but without passing the real value to the AI.
- Hide table: complete tables that are not needed for the intended analyses.
Note: masking limits what the AI can see, but offers no absolute guarantee and is not a replacement for access control. Access rights to your Power BI data are managed by Power BI itself.
Next step
Go to Add skills to assign analysis modules to your dataset.