The rapid adoption of generative artificial intelligence (GenAI) has transformed digital communication and created new challenges concerning authorship, authenticity, and trust. Conventional AI-text detection approaches primarily examine the linguistic properties of completed texts, while providing limited evidence about how a text was produced. This study proposes a multi-layer human-authorship verification framework that integrates writing-behavior signals, linguistic characteristics, and external AI-detection signals. The proposed framework, implemented through the Creative Alibi prototype, records process-oriented indicators including typing rhythm, pauses, writing bursts, revisions, deletions, and copy-paste activity, while simultaneously extracting linguistic features such as moving-average type-token ratio, sentence burstiness, hapax legomena, sentence-starter diversity, and connective-word usage. External AI detectors are incorporated as a complementary rather than definitive source of evidence. The study is designed as an empirical comparative evaluation involving human-authored, AI-generated, and AI-assisted texts. Seven experimental configurations are proposed to examine the contribution of each analytical layer and their combinations. Performance will be evaluated using accuracy, precision, recall, F1-score, ROC-AUC, confusion matrices, and participant-level validation. The framework addresses a central limitation of product-oriented AI authorship detection by treating authorship verification as a process-oriented communication problem. The study contributes to emerging research on AI-mediated communication by proposing an evidence-based approach to authenticity that considers both the final linguistic product and the process through which communication is produced.
References
Dik, S., Erdem, O., & Dik, M. (2025). Assessing GPTZero’s accuracy in identifying AI vs. human-written essays.
Kirk, C. P., & Givi, J. (2025). The AI-authorship effect: Understanding authenticity, moral disgust, and consumer responses to AI-generated marketing communications. Journal of Business Research, 186, 114984. (ScienceDirect)
Yao, G., & Liu, Z. (2025). Can AI simulate or emulate human stance? Using metadiscourse to compare GPT-generated and human-authored academic book reviews. Journal of Pragmatics, 247, 103–115. (ScienceDirect)
“AI, communication aids and the challenge of authentic authorship – Whose line is it anyway‽” (2025). Journal of Enabling Technologies, 19(2), 102–112. (ScienceDirect)
“The AI penalty and disclosure paradox: Trust, authenticity and knowledge uptake in AI-mediated communication.” (2026). Computers in Human Behavior: Artificial Humans, 8, 100304. (ScienceDirect)
“Writing with AI boosts trust-building efficiency.” (2025). iScience. (ScienceDirect)
“Combining style and semantics for robust authorship verification.” (2025). Machine Learning with Applications, 22, 100732. (ScienceDirect)
“Disentangling copy typing and natural writing behaviors using keystroke logs and deep learning models.” (2026). Assessing Writing, 69, 101070. (ScienceDirect)