Locked-Screen AI Task Output With Sensitivity-Based Privacy Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic devices with locked screens are unable to execute tasks that require generative AI models due to security concerns, preventing the output of sensitive user data, thus limiting user convenience and task processing needs.
Innovation Solution
A task processing method that allows electronic devices in a locked state to invoke a target model, ensuring output results align with a lower security level by applying security policies such as filtering, anonymization, and differential privacy to protect sensitive information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the electronic device executes task processing in a locked screen state, then user convenience and task processing capability are improved, but security risk increases due to potential privacy leakage
Solution Approach 1:
The system applies different security policies to different output contents based on their sensitivity levels. High-sensitivity contents (personal information, private messages, photos) are filtered or anonymized, while low-sensitivity contents (weather, news, public information) are output normally. This localized differentiation resolves the contradiction by allowing convenient task execution while protecting specific sensitive data.
Solution Approach 2:
The output content is segmented into multiple categories (high sensitivity, medium sensitivity, low sensitivity) with different security policies applied to each. This segmentation enables the system to process tasks conveniently while selectively protecting only the sensitive portions, rather than blocking all outputs or applying uniform security measures.
2Productivity
If the electronic device outputs all task results in locked state, then task processing completeness is improved, but privacy protection deteriorates
Solution Approach 1:
Different security policies are applied to different types of output content based on their sensitivity. Personal information and private data undergo filtering or anonymization processing, while non-sensitive task results are output completely. This resolves the contradiction by maintaining task processing completeness for non-sensitive content while protecting privacy for sensitive content.
Solution Approach 2:
A security policy module acts as an intermediary between the task processing system and the output display. This intermediary filters, anonymizes, or blocks sensitive information while allowing non-sensitive information to pass through unchanged, thus maintaining task completeness while protecting privacy.
3Reliability
If the electronic device applies strict security policies to all outputs, then privacy protection is improved, but task processing utility deteriorates
Solution Approach 1:
The system applies security policies locally to specific sensitive contents rather than uniformly to all outputs. Non-sensitive task results (weather forecasts, news summaries, public information) are output without restriction, maintaining task utility, while only sensitive personal information undergoes privacy protection processing.
Solution Approach 2:
Instead of applying security policies to all outputs (excessive action), the system applies them only to sensitive portions (partial action). This selective approach maintains task processing utility for non-sensitive content while providing privacy protection where needed, avoiding the deterioration of overall task utility.
Data Source
AI summary
A task processing method includes obtaining a target task when an electronic device is in a first operating state and a target application is active, where the target application calls a target model to respond to the target task; and in response to the target task, outputting an output result for the target task by the target application. The output result corresponds to a security level of the electronic device in the first operating state. The security level of the electronic device in the first operating state is lower than a security level of the electronic device in a second operating state.