Cybertection Compliance Auditor Ai (for Windows and Linux)

Lisense Agreement

By downloading, installing, or using the Cybertection AI software you agree to abide by the terms of this license agreement.

• Grant of License: Cybertection grants you a non-exclusive, non-transferable license to use this software for personal or professional use, in accordance with the terms and conditions outlined herein.
• Prohibited Actions: You may not:
  • Modify, reverse-engineer, decompile, or disassemble the software.
  • Distribute, sublicense, or sell the software to third parties without explicit permission.
• Ownership Rights: All intellectual property rights and ownership of the Cybertection AI software remain solely with Cybertection. This license does not convey any ownership rights.
• Liability Disclaimer: Cybertection is not responsible for any damages resulting from the use or inability to use the software. Users are advised to operate the software at their own risk.
• Termination: This license agreement is effective until terminated. Cybertection reserves the right to terminate your access if you violate any terms.
• Updates & Support: Access to updates and support may be provided at Cybertection's discretion. No guarantees are made regarding the frequency or availability of updates.
• For any questions or to request additional permissions, contact Cybertection support at cybertection@cybertection.net

The CyberTection AI Compliance Auditor AI is a desktop application with a "cyberpunk" themed graphical user interface (GUI) built using Tkinter. Its primary goal is to:

• Scan Files: Look through files in user-specified directories.
• Identify Sensitive Data: Use predefined and custom rules to find potential instances of passwords, credentials, Personal Identifiable Information (PII), Protected Health Information (PHI), financial data, etc.
• Detect Potential Issues:
  • Find files with names suggesting unencrypted sensitive content (e.g., passwords.txt).
  • Identify plaintext passwords or API keys within files.
  • Detect specific patterns like SSNs, credit card numbers, emails.
  • Check for old backup files that might violate data retention policies.
  • Look for and do basic analysis of privacy policy documents.
• Leverage AI:
  • Uses a simple built-in NLP for basic entity recognition (PERSON, ORG, DATE, TIME).
  • Can use a local Hugging Face model (distilbert-base-uncased-finetuned-sst-2-english) for text classification (e.g., sensitive vs. non-sensitive).
  • Can connect to OpenAI (GPT-3.5/GPT-4) for more nuanced analysis of text snippets for compliance risks if configured with an API key.
  • Learns from user feedback to improve accuracy (marking findings as correct hits or false positives adjusts category weights and builds a list of false positive markers).
• Report Findings: Display issues with severity levels (Critical, High, Medium, Low, Info), details, and recommendations.
• Provide General Advice: Offers general GDPR/CCPA best practice recommendations.

Key Features:

• Multi-Layered Scanning: Checks filenames, file content (using regex and AI), backup ages, and privacy policy presence.
• AI-Powered Analysis:
  • Local AI: Basic NLP and Hugging Face model for on-device processing.
  • Cloud AI (Optional): OpenAI integration for deeper contextual understanding.
• Customizable AI Model:
  • Users can add custom regex patterns.
  • Users can mark specific text snippets as false positives to be ignored.
  • Category weights (how important different types of findings are) can be adjusted through feedback.
• User Feedback Loop: Allows users to confirm if a finding is a "Correct Hit" or a "False Positive," which helps train the AI model.
• Configuration Management: AI settings (API keys, model choices, thresholds, scan limits) and the AI model's learned state can be saved and loaded.
• Severity Categorization: Helps prioritize which findings to address first.
• Non-Blocking Scans: Scanning runs in a separate thread so the UI remains responsive.
• Cyberpunk GUI: Aesthetically distinct user interface.

• • Configuration Files (Optional but Recommended):
    • cybertection_ai_config.json: Stores AI settings. If not present, defaults will be used.
    • cybertection_ai_model.json: Stores learned AI patterns and false positives. If not present, it starts fresh.

• Launch the Application:

  • Save the code as a Python file (e.g., cybertection_auditor.py).
  • Run it from your terminal: python cybertection_auditor.py

• Initial Configuration (Important for AI Features):

  • Go to the "AI Model" menu -> "Configure AI Settings".
  • Enable Local AI: Check this for basic pattern matching and local NLP/Hugging Face model usage.
  • Enable Cloud AI (OpenAI): If you want to use OpenAI:
    • Check the box.
  • Minimum Confidence Threshold: Set how confident the AI needs to be to report a finding (default 0.7).
  • Max Files to Scan / Max File Size: Adjust these to manage scan time and resource usage, especially for large directories.
  • Data Retention Days: For the "old backups" check.
  • Click "Save Configuration".

• Select Scan Directory:

  • Click the "Select Scan Directory" button.
  • Browse to and choose the folder you want to audit. The selected path will appear next to the button.

• Start the Scan:

  • Click the "START SCAN" button.
  • The "Status" bar at the bottom will show progress updates (e.g., "Phase 1: Scanning...", "Scanning file X/Y...").
  • The button will change to "SCANNING..." and become disabled until the scan is complete.

• Review Results:

  • Findings will appear in the main text area.
  • Each finding includes:
    • CHECK: The type of issue.
    • Severity: Critical, High, Medium, Low, or Info.
    • Details: Specifics about the finding, including file path and line number if applicable.
    • Recommendation: Suggested actions.
    • Category: Type of finding (e.g., "Passwords/Credentials", "PII", "AI Analysis").
  • Scroll through the results. The most critical items usually appear first or are highlighted.

• Provide AI Feedback (Crucial for Improvement):

  • Click on a specific finding in the results area. The "AI Feedback Panel" on the right will update.
  • Selected Finding: Shows a snippet of the selected finding.
  • Confidence: If the finding was AI-generated, its confidence score is shown.
  • Buttons:
    • "Correct Hit 👍": If the finding is accurate and relevant. This can boost the weight of that category or pattern.
    • "False Positive 👎": If the finding is incorrect. This can reduce the weight of the category/pattern and potentially add the snippet to a list of "false positive markers" to be ignored in the future.
  • Providing feedback helps the AI learn and become more accurate over time.

• Refine the AI Model (Advanced):

  • "AI Model" Menu:
    • "Add Custom Pattern": If you have specific internal codes, project names, or other sensitive strings not covered by default, you can add them as regex patterns with a name and category.
    • "Add False Positive Marker": If you notice recurring false positives that aren't easily tied to a single finding, you can add common non-sensitive text snippets here.
    • "View Model Statistics": See how many learned patterns, false positive markers, and the current category weights.

• Manage AI Model State:

  • "AI Model" Menu:
    • "Save AI Model State": Saves the current learned patterns, false positive markers, and category weights to cybertection_ai_model.json. Do this regularly, especially after providing feedback.
    • "Load AI Model State": Loads a previously saved model.

• Consult Help:

  • "Help" Menu:
    • "User Guide": Basic instructions (likely similar to this explanation).
    • "About": Information about the application.

• Exiting the Application:

  • Close the window. You'll be prompted to save the AI model state if you haven't recently.