Adaptive window screening for large text content security

Abstract

Systems and methods are disclosed for detecting and remediating safety issues in electronic text received by an electronic platform. The electronic text is screened using an adaptive text window that splits the text into segments. Each segment is tested for safety issues, e.g., for content moderation or computer security. Any segment that is identified as containing a potential safety issue is iteratively split and tested using progressively smaller window sizes. The testing of the segments includes generating a safety score, e.g., using a content safety model, which is compared to an adaptive threshold. The threshold may be adjusted in each iteration based on the change in window size. The dynamic screening approach continues until all segments are determined to be safe or until a minimum window size is reached and the electronic text is accepted or rejected based on the safety scores for the segments of the electronic text.