Skip to content

privacy.md

Latest commit

 

History

History
156 lines (110 loc) · 3.45 KB

privacy.md

File metadata and controls

156 lines (110 loc) · 3.45 KB

openadapt-privacy

PII/PHI scrubbing for captured demonstrations.

Repository: OpenAdaptAI/openadapt-privacy

Installation

pip install openadapt[privacy]
# or
pip install openadapt-privacy

Overview

The privacy package provides:

  • Detection of personally identifiable information (PII)
  • Detection of protected health information (PHI)
  • Redaction and anonymization
  • Compliance with privacy regulations

Detected Information Types

PII (Personally Identifiable Information)

  • Names
  • Email addresses
  • Phone numbers
  • Social Security numbers
  • Credit card numbers
  • Addresses
  • Dates of birth

PHI (Protected Health Information)

  • Medical record numbers
  • Health plan IDs
  • Account numbers
  • Certificate/license numbers
  • Medical conditions
  • Treatment information

CLI Commands

Scrub a Demonstration

openadapt privacy scrub my-task

Options:

  • --output - Output directory (default: scrubbed/)
  • --mode - Redaction mode (blur, mask, replace)
  • --types - Information types to scrub (default: all)

Preview Detection

openadapt privacy detect my-task

Shows detected PII/PHI without modifying files.

Scrub a Single Image

openadapt privacy scrub-image screenshot.png --output clean.png

Python API

from openadapt_privacy import Scrubber, PIIDetector

# Create a scrubber
scrubber = Scrubber(mode="blur")

# Scrub a demonstration
scrubber.scrub_demonstration("my-task", output_dir="scrubbed/")

# Or scrub individual images
scrubbed_image = scrubber.scrub_image(screenshot_path)

# Just detect without scrubbing
detector = PIIDetector()
detections = detector.detect(screenshot_path)

for detection in detections:
    print(f"{detection.type}: {detection.text} at {detection.bbox}")

Integration with Capture

from openadapt_capture import CaptureSession, Recorder
from openadapt_privacy import Scrubber

# Record with automatic scrubbing
session = CaptureSession(
    name="my-task",
    scrubber=Scrubber(mode="blur")
)

recorder = Recorder(session)
recorder.start()
# ... demonstration collection ...
recorder.stop()

# Demonstrations are automatically scrubbed

Redaction Modes

Mode Description Use Case
blur Gaussian blur over sensitive areas Visual redaction
mask Black box over sensitive areas Complete hiding
replace Replace with placeholder text Maintaining layout

Key Exports

Export Description
Scrubber Main scrubbing class
PIIDetector PII detection
PHIDetector PHI detection
Detection Detection result
RedactionMode Redaction options

Detection Models

Model Types Accuracy
presidio PII High
philter PHI High
regex Common patterns Medium
custom User-defined -

Compliance

This package helps with compliance for:

  • GDPR (General Data Protection Regulation)
  • HIPAA (Health Insurance Portability and Accountability Act)
  • CCPA (California Consumer Privacy Act)

!!! warning "Disclaimer" This tool assists with privacy protection but does not guarantee compliance. Always consult with legal and compliance experts for your specific use case.

Related Packages