AI-Powered Redaction

SAR Portal uses artificial intelligence to automatically detect personal data (PII) in documents and help you redact it before responding to data subjects.

How AI Redaction Works

Two-Layer Detection

SAR Portal uses a dual-layer AI system for maximum accuracy:

Layer 1: Pattern Analysis

Detects 50+ entity types
Recognizes structural patterns
Identifies common PII formats
Fast initial detection

Layer 2: Contextual AI Analysis

Contextual understanding
Data subject identification
Complex pattern recognition
Nuanced entity detection

The Process

Document Analysis - AI extracts text from the document
PII Detection - Both layers scan for personal data
Entity Mapping - Detected items are mapped to document coordinates
Review - You review and confirm detected entities
Redaction - Confirmed items are redacted
Output - Clean document ready for delivery

Starting AI Analysis

Single Document Analysis

Open the case with the document
Click Analyze for PII on the document
Wait for analysis to complete (typically 10-30 seconds)
Review detected entities

Batch Analysis

For multiple documents:

Select documents to analyze
Click Batch Analyze
Monitor progress
Review results for each document

Understanding Detection Results

Entity Types Detected

Category	Examples
Personal Identifiers	Names, email addresses, phone numbers
Government IDs	Passport numbers, national IDs, SSN/PPS
Financial	Bank accounts, credit card numbers, IBAN
Location	Addresses, postcodes, GPS coordinates
Health	Medical conditions, prescriptions, diagnoses
Professional	Job titles, employee IDs, company names
Digital	IP addresses, usernames, device IDs
Dates	Birth dates, event dates, timestamps

Confidence Levels

Each detection includes a confidence score:

High (90%+) - Very likely PII, recommend redacting
Medium (70-90%) - Probable PII, review carefully
Low (<70%) - Possible PII, manual verification needed

Reviewing Detections

The Review Interface

After analysis, you’ll see:

Document preview with highlighted entities
List of detected items
Entity type and confidence for each
Options to confirm, reject, or edit

Actions for Each Detection

Confirm - Accept the detection and include in redaction

Reject - Remove from redaction list (not PII or needed in response)

Edit - Adjust the selection boundaries

Adding Manual Detections

If the AI missed something:

Use the selection tool
Highlight the text
Specify the entity type
Add to redaction list

Applying Redactions

Preview Before Applying

Always preview redactions before applying:

See exactly what will be removed
Verify no important data is lost
Ensure completeness

Redaction Methods

Black Box - Covers text with solid black rectangle

White Box - Covers text with white (for light backgrounds)

Permanent Redaction

Important

Redactions are permanent and cannot be undone. The original document is preserved, and a new redacted version is created.

Best Practices

Review All Detections

Don’t blindly accept all AI detections:

Check context - is this the data subject’s information?
Verify third-party data is redacted
Ensure nothing required for the response is removed

Protect Third Parties

When responding to access requests:

Redact other people’s personal data
Redact information that would identify others
Keep only the requesting subject’s data

Document Your Decisions

Note why certain detections were rejected
Record manual additions
Create audit trail for compliance

Use Appropriate Formats

For best AI results:

Use searchable PDFs (not scanned images when possible)
Ensure text is selectable
Higher quality = better detection

AI Quotas

AI features are quota-based:

Plan	Monthly AI Budget
Trial	~1,000 operations
Basic	Tier-specific limit
Starter	Higher limits
Pro	Highest limits

What Counts Toward Quota

Document text extraction
PII detection analysis
AI recommendations

Monitoring Usage

View your AI usage on:

Dashboard usage widget
Billing page
Per-document cost indicators

Supported Document Types

AI redaction works with over 28 file types:

Format	Text Extraction	Redaction Method
PDF (searchable)	Yes	Visual redaction
PDF (scanned)	Yes (OCR)	Visual redaction
Word (.docx, .dotx, .docm, .dotm)	Yes	Visual redaction
Excel (.xlsx, .xlsm, .xltx, .xltm)	Yes	Cell-level redaction
Images (.png, .jpg, .gif, .bmp, .tiff, .webp)	Yes (OCR)	Visual redaction
Email (.eml, .msg)	Yes (headers + body)	PDF output
Text (.txt, .csv, .log, .md, .json, .xml, .html, .css, .js)	Yes	Text replacement

AI Risk Assessment

SAR Portal includes AI-powered risk assessment to help you identify cases that may require special attention.

How Risk Assessment Works

The AI analyzes case details and documents to identify:

Sensitive data categories - Health, financial, or other high-risk data
Complexity indicators - Multiple systems, large document sets
Compliance risks - Deadline concerns, incomplete information
Third-party data - Data about others that needs protection

Risk Levels

Level	Description	Recommended Action
Low	Standard request, minimal sensitive data	Normal processing
Medium	Some complexity or sensitive data present	Careful review
High	Complex request or highly sensitive data	Senior review recommended

Using Risk Assessment

Open a case with documents
Click Assess Risk or run during AI analysis
Review the risk summary
Follow recommendations for your risk level

Benefits

Prioritize high-risk cases appropriately
Identify cases needing legal review
Ensure proper handling of sensitive data
Document risk-based decisions

Troubleshooting

“No PII Detected”

Document may not contain personal data
Check if document has extractable text
Scanned documents may need better quality

Low Detection Accuracy

Verify document quality
Check language support
Consider manual review for edge cases

Analysis Taking Too Long

Large documents take longer
Complex layouts require more processing
Check document size limits