Abstract:
In today's digital world, across communication platforms, authentication systems, and digital services, a large volume of data is typed. Traditional key loggers, that do not filter and capture every key stroke, produce an excessive amount of irrelevant data that makes it nearly impossible to identify sensitive information. To tackle this problem, this study presents a progressive approach by integrating artificial intelligence into a MASM x86 low-level assembly key logger. The suggested system facilitates precise differentiation between ordinary text and sensitive data such as user names and passwords. The system achieves this through the analysis of key stroke sequences and pin pointing outliers in character structures. While standard text generally recognizes patterns composed of alphabetic or numeric characters, sensitive data in most cases exhibits irregular arrangements of alphabetic, numeric, and special symbols. By exploiting these distinctive features, the AI models minimize redundant logging,emphasize potentially critical key strokes, and enhance the overall efficiency of capturing. The results highlight the feasibility of merging AI-driven classification with low-level monitoring tools. Still, the system has some limitations, depending on the quality of the training dataset, and them is understanding of weak or short passwords. Moreover, ethical and legal consequences limit the practical deployment of these systems to controlled research environments. Regardless of these restrictions, this research opens the path for more advanced, selective, and efficient key stroke analysis methodologies.
Page(s):
118-118
DOI:
DOI not available
Published:
Journal: 4th International Conference of Sciences “Revamped Scientific Outlook of 21st Century, 2025” , November 12,2025, Volume: 1, Issue: 1, Year: 2025
Keywords:
Artificial Intelligence
,
sensitive data detection
,
Keylogger
,
MASM x86 assembly
,
keystroke analysis