Part ofProtection and Enforcement
AI-Assisted Data Masking
Generate Context-Aware Replacement Values With a Local LLM
Use a locally configured language model to create fictitious values for dynamic masking, static masking, and event-tagging workflows.
- Local Model
- Operator-Written Prompts
- Dynamic and Static Masking
Local LLM masking
Generate realistic replacements inside controlled masking rules
DataSunrise can use a locally configured language model to generate context-aware fictitious values for masking. Teams control the model, prompt, fields, and rule.
The AI Generated Value method works inside DataSunrise Dynamic Masking, Static Masking, and event-tagging workflows. It is designed for values that need more context than a fixed string, random number, or simple dictionary can provide.
AI-assisted masking workflow
Generate a replacement inside a controlled rule
A locally configured model adds context-aware value generation to DataSunrise masking workflows.
Configured masking rule
- Selected source and field
- Operator-written masking prompt
- Dynamic, static, or tagging workflow
Local LLM integration
- Configured model file
- Context, temperature, seed, and timeout
- Tested output before activation
Fictitious replacement
- Generated value returned to the rule
- Original policy remains in control
- Result used in the selected workflow
Generate Useful Fictitious Values
A masking prompt tells the local model what kind of replacement to create. DataSunrise sends the selected value or value batch with that prompt and returns the generated result through the configured masking workflow.
This gives teams another masking method for text and business data where a realistic replacement is more useful than a constant mask. The generated value remains governed by the rule that selected the source, field, and user context.
Use the Method Across DataSunrise Workflows
| Workflow | How AI Generated Value is used |
|---|---|
| Dynamic Data Masking | Generate a replacement for a live masking result |
| Static Data Masking | Generate fictitious values while DataSunrise builds a protected target copy |
| Event Tagging | Generate values for configured data-classification and tagging workflows |
The local model provides the value. DataSunrise decides when to call it through the configured rule and masking conditions.
Configure and Test the Local Model
Administrators add the model file to DataSunrise and configure its context size, temperature, processing threads, seed, and test timeout. A built-in test sends sample data with a masking prompt so the team can review the output before the model is used in a rule.
After testing, the operator selects the model and prompt in the AI Generated Value masking method. A response timeout keeps the masking workflow within the performance limit chosen for the deployment.
Use AI When a Conventional Mask Is Too Rigid
AI Generated Value is useful when a replacement needs to fit the surrounding business context. Teams can test the output, tune the prompt and model settings, and set a timeout before using the method in a live or batch workflow.
The model generates the replacement value. DataSunrise rules still determine where and when the method runs.
Frequently Asked Questions
Does AI-Assisted Data Masking send values to ChatGPT or another public AI service?
No. This product uses a model file configured through the DataSunrise local LLM integration.
Can the model decide what data should be masked?
No. The operator selects the rule, source, fields, model, and prompt. The model generates the replacement value inside that configuration.
Is this Generative AI traffic protection?
No. Generative AI Data Masking and Prompt Controls protects traffic sent to named AI services. AI-Assisted Data Masking uses a local model as a masking method for DataSunrise data workflows.
Map the right workflow