Real-Time Image and Audio Anonymization in Secure Sandboxes

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing augmented reality (AR) and virtual reality (VR) systems fail to effectively preserve user privacy by inadvertently capturing and sharing sensitive information such as credit card statements or private conversations.

Innovation Solution

Implementing a secure data vault system with generative AI models that process raw data from cameras and microphones within a sandbox environment to detect and obfuscate private information in real-time, ensuring privacy by blurring or muting sensitive content.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If AR/VR systems capture media content from cameras and microphones, then augmented reality visualizations and communication functionality are enabled, but private information such as credit card statements and private conversations are inadvertently captured and shared

Engineering Contradiction:

ImproveAR/VR functionalityVSAvoidprivacy violation

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

The system performs preliminary action by detecting and identifying private information in media content before it is shared or processed further. The machine learning model analyzes captured images and audio in real-time to detect sensitive content such as credit cards, licenses, and private conversations, allowing the system to take protective action before privacy violations occur.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system converts the harmful effect of inadvertently capturing private information into a beneficial privacy protection mechanism. By using machine learning to detect private information and automatically applying obfuscation effects (blurring images, muting audio), the system transforms potential privacy violations into enhanced privacy protection, allowing AR/VR functionality to continue while safeguarding user data.

Inventive Principle:

Principle #22Blessing in disguise (Convert harm into benefit)

2Reliability

If real-time detection and obfuscation of private information is implemented, then privacy protection is enhanced, but processing time and computational resources are increased

Engineering Contradiction:

Improveprivacy protectionVSAvoidprocessing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system applies partial action by selectively processing only the portions of media content that are likely to contain private information. Rather than processing every pixel and audio sample uniformly, the machine learning model identifies regions of interest (such as documents or specific audio patterns) and applies obfuscation only to those areas, reducing overall processing time while maintaining effective privacy protection.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The system maintains continuity of useful action by performing detection and obfuscation operations continuously in real-time as media content is captured, rather than processing in batches. This continuous processing allows the system to protect privacy without significant delays to AR/VR functionality, as the obfuscation is applied immediately when private information is detected during the capture process.

Inventive Principle:

Principle #20Continuity of useful action

3Measurement precision

If machine learning models are used to detect private information, then detection accuracy is improved, but the complexity of the system increases

Engineering Contradiction:

Improvedetection accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The machine learning model serves as an intermediary component between the camera/microphone capture system and the final media output. This intermediary layer automatically detects private information and triggers appropriate obfuscation effects, simplifying the overall system architecture by centralizing the detection and protection logic in a dedicated module rather than requiring complex integration throughout the entire AR/VR system.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS12450402B2Real-time anonymization of images and audio

Publication Date: 2025.10.21 SNAP INC

AI summary

Systems and methods are provided. A system includes a display and camera. The system additionally includes a secure data vault system. The secure data vault system includes a sandbox system operatively coupled to the camera and configured to receive camera data from the camera, wherein in operation of the sandbox system, the camera only sends camera data to the sandbox system, and wherein the sandbox system comprises an execution environment configured to restrict execution of instructions to a predefined memory address range. The secure data vault system additionally includes a display and rendering system operatively coupled to the sandbox system and configured to render an image based on the camera data processed via the instructions and to display the image via the display, wherein the display and rendering system is configured to blur sections of the image based on private information derived from the image.