Voice Assistant Audio Redaction for Enterprise Data Security
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In enterprise settings, personal voice assistants often audibly announce sensitive enterprise data in insecure environments, risking data leaks due to the lack of redaction mechanisms when multiple individuals are present.
Innovation Solution
A computing environment connects voice assistant devices to a remote service that generates audio signals for enterprise data, with redaction capabilities based on user and device contexts, automatically switching to higher security modes when multiple individuals are detected, using predefined policies and filters to replace sensitive data with predetermined tones or silence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If voice assistant devices audibly announce enterprise data verbatim, then information accessibility is improved, but data security deteriorates due to potential overhearing by unauthorized individuals
Solution Approach 1:
The patent applies local quality by making different parts of the spoken output have different properties - some portions are spoken verbatim while sensitive portions are redacted or replaced with tones. This allows the voice assistant to maintain accessibility for non-sensitive information while protecting sensitive information, resolving the contradiction between information accessibility and data security.
Solution Approach 2:
The system performs preliminary action by analyzing the context (number of people present, device type, user role) before generating the audio output. This allows the system to pre-determine which portions of information should be redacted before speaking, preventing data leaks while maintaining accessibility for appropriate information.
2Object-affected harmful factors
If context-aware redaction mechanisms are implemented, then data security is improved, but device complexity increases due to additional processing requirements
Solution Approach 1:
The patent introduces an intermediary component (the computing environment or service) that handles the complex redaction processing. Instead of making the voice assistant device itself complex, the system offloads the analytical and redaction processing to a separate service that the device communicates with, thereby improving data security without significantly increasing device complexity.
3Object-affected harmful factors
If security modes are automatically switched based on detected individuals, then data security is improved, but ease of operation deteriorates due to reduced user control
Solution Approach 1:
The system implements feedback by continuously monitoring the environment (detecting number of people present) and automatically adjusting the security level accordingly. When multiple people are detected, the system feedback-loop adjusts to a higher security mode with more redaction. This automated feedback mechanism improves data security while maintaining ease of operation by removing the need for manual user intervention.
Data Source
AI summary
Disclosed are various examples for audio data leak prevention using user and device contexts. In some examples, a voice assistant device can be connected to a remote service that provides enterprise data to be audibly emitted by the voice assistant device. In response to a request for the enterprise data being received from the voice assistant device, an audio signal can be generated that audibly broadcasts the enterprise data. The audio signal can be generated to audibly redact at least a portion of the enterprise data based at least in part on a mode of operation of the voice assistant device. The voice assistant device can be directed to emit the enterprise data through a playback of the audio signal.