From static to dynamic risk indicators in predicting and detecting insider attacks
Bibliographic record
Abstract
Cyber insider threats represent one of the most complex and insidious challenges to modern cybersecurity, as they originate from legitimate users whose behaviors may turn malicious over time. Traditional approaches often fail due to their reliance on static risk indicators, necessitating dynamic modeling of human behavior to capture evolving risks. In this paper, we propose a novel framework for detecting insider threats by continuously and dynamically inferring personality-based risk indicators from employees’ writing data using a publicly accessible large language model (Meta AI’s Llama-3.2). These indicators are modeled as time series and processed through AutoRegressive Integrated Moving Average (ARIMA) models to forecast behavioral deviations. Predicted anomalies are subsequently classified using a hybrid ensemble combining Artificial Neural Networks (ANN) and Random Forest (RF) to distinguish benign variations from genuine insider threats. Our framework identifies behavioral anomalies, provides interpretable detection windows, and achieves the following results on CMU-CERT datasets (r4.2/r5.2): recall (90.0%/86.0%), precision (92.6%/87.7%), ROC-AUC (94.7%/92.6%), MSE (0.231/0.304), MTTD (21.3 days/31.72 days) and a median latency under 230 ms for real-time operation. These results demonstrate significant improvements over baseline static approaches in both detection accuracy and temporal prediction capability. This work advances human-centric and proactive insider threat detection by integrating personality-based risk indicators with predictive modeling, providing a scalable and interpretable solution for real-time dynamic risk assessment in enterprise environments.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".