Proactive Cyber Threat Detection Using AI and Open-Source Intelligence
Bibliographic record
Abstract
Frequent developments in cyber threats seriously threaten the digital systems in both the public and private sectors. Today, modern cyberattacks are too unpredictable for the old cybersecurity defenses and time-bound detection methods. Because there are more complex, numerous and distant threats today, to find them and address them before much damage can occur. In this work, look at integrating AI and OSINT to develop a system that can quickly detect any cyber threats in an organization. The researchers used the Hornet 40 dataset which includes network traffic collected over the course of 40 days from honeypots in eight places: Amsterdam, London, Frankfurt, San Francisco, New York, Singapore, Toronto, and Bangalore. To capture different activities from uninvited users, these honeypots received requests only on a specific non-standard SSH port. The information provided by Argus is in the form of detailed bidirectional NetFlow data that displays the effects of geography on various cyber-attacks. Various machine learning approaches are used within a data-driven system to spot and detect abnormal traffic and threats in the network such as Random Forest, Support Vector Machines (SVM), Long Short-Term Memory (LSTM) networks and Isolation Forests. At the same time, data, and findings from public threat intelligence, darknet sources and cybersecurity forums are studied using Natural Language Processing (NLP) to find important information about threats. As a result of this, the detection rate is improved by comparing suspicious traffic in honeypots with global findings and the reported IOCs. Combining AI and OSINT together allows the engine to read and analyze a lot of network data quickly and in almost real time. Joining these processes allows quick and early identification of advanced attacks such as zero-day attacks and intrusions. It is clear from the results that using this approach improves the accuracy of detection, lowers the number of false positives, and reveals attacks that tend to come from specific locations and are typically overlooked by other systems.
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 | 0.001 | 0.001 |
| Scholarly communication | 0.000 | 0.002 |
| Open science | 0.001 | 0.002 |
| 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".