By: Arya Brijith, International Center for AI and Cyber Security Research and Innovations (CCRI), Asia University, Taiwan, firstname.lastname@example.org
For telecom firms, the surge in fraudulent activities has become a serious problem that affects both their revenue and the overall quality of their services. The basic necessity for reliable fraud detection algorithms is highlighted in this paper. Deceptive tactics like Social Engineering, (a manipulative technique that cybercriminals and scammers mostly use to perform such actions or make decisions that are not in their best interests.) require a comprehensive approach that integrates technological solutions such as the CNN algorithm, alongside public awareness, regulatory measures, and global cooperation. Furthermore, this article places a strong emphasis on the pivotal role of Artificial Intelligence (AI) in the battle against scam calls within the dynamic telecommunications arena. While scams may persist and adapt, the collaborative effort to leverage the potential of AI and other cutting-edge tools stands as a robust defence against this contemporary threat. The ongoing strides in technological advancements bring us closer to a future devoid of scam calls, highlighting the collective resolve to safeguard individuals and society from this pervasive danger.
The growing incidence of fraud worries telecom businesses. Technology and system information advancements have greatly enhanced fraud activity, which can have an adverse impact on revenue and service quality. Scammer often use a variety of techniques and platforms to elude authorities. Therefore, it is crucial for telecom companies to create effective algorithms that can spot possible scams before they occur [8-11]. The process includes data collection, data visualization, feature extraction, training and testing the data, creating evaluation metrics and then deploying the model. Obtaining data for the study of scam calls is challenging, as true scam victims do not in general record their conversations.  As the dataset must include real-time conversations between the scammer and the victim, we can use resources such as YouTube channels(example: ScamBaiters).The model must be trained to detect scam calls and provide an alert message to the user in order to prevent any fraudulent activities from occurring. Further, let’s discuss the effects of scamming and how AI techniques can be implemented to prevent scams.
Effects of scamming
To rob individuals, con artists employ a number of techniques. Scams offering “miracle” health cures or “get rich quick” opportunities often use fake testimonials from “satisfied customers” to induce trust. The recipient of the scam message is likely to overestimate the reliability and validity of the scam message if it has the apparent backing of others, as well as overestimating the trustworthiness of the scam source. This undermines confidence not only between people but also between people and reliable businesses or institutions. Additionally, fraudsters use new communication channels and strategies as technology advances, making it challenging for law enforcement to keep on top of their strategies.
The scammers’ attempts to establish similarity with their victims are also an example of this kind of social influence attempt.. It requires taking advantage of psychological and emotional elements like trust, authority, fear, or curiosity in order to compel someone into disclosing sensitive information or acting in ways that jeopardize their security.
Strategies and Algorithms to prevent scam calls
- AI-Based Cyber security Systems: AI can play a significant role in preventing scam calls and protecting individuals from falling victim to fraudulent activities as today’s use of AI in cyber security has several advantages. It has automated features which act at a fast response time, ensuring safety. The systems are therefore able to detect key threats and create ways through which these attacks could be avoided.. It can detect new attacks humans don’t understand.
- VoIP spam detection: VoIP spam detection process does not pertain to a single technique of detection. At different points, the detection must be performed utilizing an assortment of ways. Majority of spam is removed at each step by the spam detection method qualified by that stage, and any further spam that is sent through or forwarded would be quarantined at the following stage. Two security protocols used in VoIP communication are secure SIP (Session Initiation Protocol) and SRTP (Secure real-time transport protocol). RTP delivers audio or video data, whereas SIP establishes the connection between peers. Anyone with a minimal understanding of network technology is capable of intercepting SIP messages and RTP streams .
- Kernel-based Online Anomaly Detection (KOAD): The Kernel-based Online Anomaly Detection (KOAD) technique makes no previous assumptions on a model of typical or abnormal network data. Instead, it gradually compiles a dictionary of characteristics that roughly covers the subspace of typical network behaviour. The dictionary is flexible and adapt to changes in the structure of everyday traffic. When the algorithm notices a departure from the norm, an alarm is then promptly raised. In order to maintain a compact dictionary, the KOAD algorithm also removes out-of-date items as the zone of routine changes or migrates.
- Identifying patterns and known scam numbers: AI-driven Natural Language Processing (NLP) algorithms can analyse the content of phone calls or messages to detect signs of scams or phishing attempts. They can identify keywords or phrases commonly used by scammers and raise red flags accordingly.
- Adaptive Learning and anomaly detection: AI systems can continuously learn from new scam patterns and adapt their algorithms to counter emerging threats effectively. It can monitor call and messaging traffic to detect unusual or anomalous behaviour. For instance, if a user suddenly receives a high volume of calls from unknown numbers, AI can flag this as suspicious activity.
With reference to the above image
Deep Learning (DL) techniques, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), can be used for more advanced voice and speech analysis to detect unusual characteristics or anomalies in callers’ voices. Context comprehension is a strong suit of deep learning. Consider this: con artists don’t just read from a script; they participate in discussions, frequently attempting to deceive targets with their words. Recurrent neural networks (RNNs) and transformers are two examples of deep learning models that can really understand the content of these talks[12-13].
In the ever-evolving landscape of telecommunications, where technology is a double-edged sword, protecting individuals and society from the pervasive threat of scam calls has become an urgent necessity. This article has shed light on the multifaceted issue of scam calls, the impact they have on people’s lives, and the critical role that artificial intelligence (AI), can play in their prevention. The strategies used in scam calls may be quite cunning, scammers constantly alter their strategy in an effort to trick the system; nonetheless, deep learning techniques are excellent at identifying new patterns and fraud.
In conclusion, while scam calls may persist and evolve, our collective determination to harness the power of AI and other tools will continue to serve as a formidable defence against this modern threat. Also, deep learning may be thought of as the best ally in the struggle against swindler calls. It is a game-changer in combating the constantly changing world of fraudulent calls because it is intelligent, flexible, and always vigilant In a nutshell, the journey towards a scam-free future is underway, and with every technological advancement, we take another step closer to achieving that goal.
- Wood, I., Kepkowski, M., Zinatullin, L., Darnley, T., & Kaafar, M. A. (2023). An analysis of scam baiting calls: Identifying and extracting scam stages and scripts. arXiv preprint arXiv:2307.01965.
- Fischer, P., Lea, S. E., & Evans, K. M. (2013). Why do individuals respond to fraudulent scam communications and lose money? The psychological determinants of scam compliance. Journal of Applied Social Psychology, 43(10), 2060-2072.
- Ansari, M. F., Sharma, P. K., & Dash, B. (2022). Prevention of phishing attacks using AI-based Cybersecurity Awareness Training. Prevention.
- Dantu, R., & Kolan, P. (2005). Detecting Spam in VoIP Networks. SRUTI, 5, 5-5.
- Y. -S. Wu, S. Bagchi, N. Singh and R. Wita, “Spam detection in voice-over-IP calls through semi-supervised clustering,” 2009 IEEE/IFIP International Conference on Dependable Systems & Networks, Lisbon, Portugal, 2009, pp. 307-316, doi: 10.1109/DSN.2009.5270323.
- Ahmed, T., Oreshkin, B., & Coates, M. (2007, April). Machine learning approaches to network anomaly detection. In Proceedings of the 2nd USENIX workshop on Tackling computer systems problems with machine learning techniques (pp. 1-6). USENIX Association.
- Casillo, M., Colace, F., Gupta, B. B., Santaniello, D., & Valentino, C. (2021). Fake news detection using LDA topic modelling and K-nearest neighbor classifier. In Computational Data and Social Networks: 10th International Conference, CSoNet 2021, Virtual Event, November 15–17, 2021, Proceedings 10 (pp. 330-339). Springer International Publishing. https://link.springer.com/chapter/10.1007/978-3-030-91434-9_29
- Sahoo, S. R., & Gupta, B. B. (2021). Multiple features based approach for automatic fake news detection on social networks using deep learning. Applied Soft Computing, 100, 106983.
- Gupta, B. B., Tewari, A., Jain, A. K., & Agrawal, D. P. (2017). Fighting against phishing attacks: state of the art and future challenges. Neural Computing and Applications, 28, 3629-3654.
- Zhang, Q., Guo, Z., Zhu, Y., Vijayakumar, P., Castiglione, A., & Gupta, B. B. (2023). A deep learning-based fast fake news detection model for cyber-physical social services. Pattern Recognition Letters, 168, 31-38.
- Casillo, M., Colace, F., Gupta, B. B., Santaniello, D., & Valentino, C. (2021). Fake news detection using LDA topic modelling and K-nearest neighbor classifier. In Computational Data and Social Networks: 10th International Conference, CSoNet 2021, Virtual Event, November 15–17, 2021, Proceedings 10 (pp. 330-339). Springer International Publishing.
- Gupta, B. B., & Sheng, Q. Z. (Eds.). (2019). Machine learning for computer and cyber security: principle, algorithms, and practices. CRC Press.
- Gupta, B. B., Perez, G. M., Agrawal, D. P., & Gupta, D. (2020). Handbook of computer networks and cyber security. Springer, 10, 978-3.
Brijith A. (2023) Unmasking Scam Calls: Analysing and Detecting Scammers using AI, Insights2Techinfo, pp.1