On the Insecurity of Keystroke-Based AI Authorship Detection: Timing-Forgery Attacks Against Motor-Signal Verification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Author: Condrey, David
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912845264846848
author Condrey, David
author_facet Condrey, David
contents Recent proposals advocate using keystroke timing signals, specifically the coefficient of variation ($δ$) of inter-keystroke intervals, to distinguish human-composed text from AI-generated content. We demonstrate that this class of defenses is insecure against two practical attack classes: the copy-type attack, in which a human transcribes LLM-generated text producing authentic motor signals, and timing-forgery attacks, in which automated agents sample inter-keystroke intervals from empirical human distributions. Using 13,000 sessions from the SBU corpus and three timing-forgery variants (histogram sampling, statistical impersonation, and generative LSTM), we show all attacks achieve $\ge$99.8% evasion rates against five classifiers. While detectors achieve AUC=1.000 against fully-automated injection, they classify $\ge$99.8% of attack samples as human with mean confidence $\ge$0.993. We formalize a non-identifiability result: when the detector observes only timing, the mutual information between features and content provenance is zero for copy-type attacks. Although composition and transcription produce statistically distinguishable motor patterns (Cohen's d=1.28), both yield $δ$ values 2-4x above detection thresholds, rendering the distinction security-irrelevant. These systems confirm a human operated the keyboard, but not whether that human originated the text. Securing provenance requires architectures that bind the writing process to semantic content.
format Preprint
id arxiv_https___arxiv_org_abs_2601_17280
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle On the Insecurity of Keystroke-Based AI Authorship Detection: Timing-Forgery Attacks Against Motor-Signal Verification
Condrey, David
Cryptography and Security
Artificial Intelligence
Human-Computer Interaction
68P30 (Primary) 94A15 (Secondary)
K.6.5; I.2.7; I.5.4
Recent proposals advocate using keystroke timing signals, specifically the coefficient of variation ($δ$) of inter-keystroke intervals, to distinguish human-composed text from AI-generated content. We demonstrate that this class of defenses is insecure against two practical attack classes: the copy-type attack, in which a human transcribes LLM-generated text producing authentic motor signals, and timing-forgery attacks, in which automated agents sample inter-keystroke intervals from empirical human distributions. Using 13,000 sessions from the SBU corpus and three timing-forgery variants (histogram sampling, statistical impersonation, and generative LSTM), we show all attacks achieve $\ge$99.8% evasion rates against five classifiers. While detectors achieve AUC=1.000 against fully-automated injection, they classify $\ge$99.8% of attack samples as human with mean confidence $\ge$0.993. We formalize a non-identifiability result: when the detector observes only timing, the mutual information between features and content provenance is zero for copy-type attacks. Although composition and transcription produce statistically distinguishable motor patterns (Cohen's d=1.28), both yield $δ$ values 2-4x above detection thresholds, rendering the distinction security-irrelevant. These systems confirm a human operated the keyboard, but not whether that human originated the text. Securing provenance requires architectures that bind the writing process to semantic content.
title On the Insecurity of Keystroke-Based AI Authorship Detection: Timing-Forgery Attacks Against Motor-Signal Verification
topic Cryptography and Security
Artificial Intelligence
Human-Computer Interaction
68P30 (Primary) 94A15 (Secondary)
K.6.5; I.2.7; I.5.4
url https://arxiv.org/abs/2601.17280