Advanced AI-Powered Phishing URL Detection Using Ensemble Algorithm with RNN, BILSTM, GRU Approach
Bibliographic record
Abstract
Phishing attacks represent a major security threat to cybersecurity, often tricking users into divulging sensitive information through malicious URLs. To combat this growing concern, our project, Advanced Phishing URL Detection Through AI Techniques, leverages deep learning-based models to classify URLs as either phishing or Authentic. Utilizing the ISCX-URL dataset from the Canadian Institute for Cybersecurity, we explore the effectiveness of various recurrent neural network (RNN) architectures for phishing URL detection. We implemented and evaluated three deep learning models—Recurrent Neural Networks (RNN), Gated Recurrent Units (GRU), and Bidirectional Long Short-Term Memory (BiLSTM). Our experimental results indicate that the BiLSTM model outperformed the others, achieving a test accuracy of 90%, followed by GRU at 87% and RNN at 84%. These findings suggest that incorporating bidirectional context awareness significantly enhances phishing detection accuracy. This research highlights the potential of AI-driven methods in strengthening cybersecurity defences against phishing attacks.In this work, we present a comparative study of deep learning models—RNN, GRU, and BiLSTM—for phishing URL detection using the ISCX-URL dataset. Our primary contribution lies in the empirical demonstration that BiLSTM, due to its bidirectional context awareness, achieves the highest test accuracy of 90.14%, outperforming GRU 87.10% and RNN 84.00%. Additionally, we apply preprocessing techniques and experiment with model tuning to enhance classification accuracy. These findings underscore the applicability of deep learning for real-time phishing detection in cybersecurity.
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 | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| 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".