Tamás Vörös
Research Subjects
Tamás Vörös has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Ben Gelman
Sophos AI Team
2 shared publications - Konstantin Berlin
Sophos Inc
2 shared publications - Salma Taoufiq
Sophos AI Team
2 shared publications - Sean Paul Bergeron
Sophos Inc
2 shared publications - Adarsh Kyadige
Sophos AI Team
1 shared publication - Sean Gallagher
Sophos AI Team
1 shared publication
Latest Publications
- Open Access
- 12 April 2024
- Published by Springer Nature
- p. 81-86
Phishing and Social Engineering in the Age of LLMs
The human factor remains a major vulnerability in cybersecurity. This chapter explores the escalating threats that Large Language Models (LLMs) pose in the field of cybercrime, particularly in phishing and social engineering. Due to their ability to generate highly convincing and individualized content, LLMs enhance the effectiveness and scale of phishing attacks, making them increasingly difficult to detect. The integration of multimodal generative models allows malicious actors to leverage AI-generated text, images, and audio, increasing attack avenues and making attacks more convincing. Two case studies provide a comprehensive look, examining how AI technology orchestrates a phishing attack posing as a typical e-commerce transaction and how an LLM was used in a romance-themed cryptocurrency scam. Both scenarios underline the need for increased awareness and improved defenses against these novel and sophisticated cyber threats.
- 12 April 2024
- Published by Springer Nature
- p. 81-86
Open AccessPhishing and Social Engineering in the Age of LLMs
The human factor remains a major vulnerability in cybersecurity. This chapter explores the escalating threats that Large Language Models (LLMs) pose in the field of cybercrime, particularly in phishing and social engineering. Due to their ability to generate highly convincing and individualized content, LLMs enhance the effectiveness and scale of phishing attacks, making them increasingly difficult to detect. The integration of multimodal generative models allows malicious actors to leverage AI-generated text, images, and audio, increasing attack avenues and making attacks more convincing. Two case studies provide a comprehensive look, examining how AI technology orchestrates a phishing attack posing as a typical e-commerce transaction and how an LLM was used in a romance-themed cryptocurrency scam. Both scenarios underline the need for increased awareness and improved defenses against these novel and sophisticated cyber threats.
- Open Access
Web Content Filtering through knowledge distillation of Large Language Models
We introduce a state-of-the-art approach for URL categorization that leverages the power of Large Language Models (LLMs) to address the primary objectives of web content filtering: safeguarding organizations from legal and ethical risks, limiting access to high-risk or suspicious websites, and fostering a secure and professional work environment. Our method utilizes LLMs to generate accurate classifications and then employs established knowledge distillation techniques to create smaller, more specialized student models tailored for web content filtering. Distillation results in a student model with a 9% accuracy rate improvement in classifying websites, sourced from customer telemetry data collected by a large security vendor, into 30 distinct content categories based on their URLs, surpassing the current state-of-the-art approach. Our student model matches the performance of the teacher LLM with 175 times less parameters, allowing the model to be used for in-line scanning of large volumes of URLs, and requires 3 orders of magnitude less manually labeled training data than the current state-of-the-art approach. Depending on the specific use case, the output generated by our approach can either be directly returned or employed as a pre-filter for more resource-intensive operations involving website images or HTML.
Open AccessWeb Content Filtering through knowledge distillation of Large Language Models
We introduce a state-of-the-art approach for URL categorization that leverages the power of Large Language Models (LLMs) to address the primary objectives of web content filtering: safeguarding organizations from legal and ethical risks, limiting access to high-risk or suspicious websites, and fostering a secure and professional work environment. Our method utilizes LLMs to generate accurate classifications and then employs established knowledge distillation techniques to create smaller, more specialized student models tailored for web content filtering. Distillation results in a student model with a 9% accuracy rate improvement in classifying websites, sourced from customer telemetry data collected by a large security vendor, into 30 distinct content categories based on their URLs, surpassing the current state-of-the-art approach. Our student model matches the performance of the teacher LLM with 175 times less parameters, allowing the model to be used for in-line scanning of large volumes of URLs, and requires 3 orders of magnitude less manually labeled training data than the current state-of-the-art approach. Depending on the specific use case, the output generated by our approach can either be directly returned or employed as a pre-filter for more resource-intensive operations involving website images or HTML.
- Open Access
That Escalated Quickly: An ML Framework for Alert Prioritization
In place of in-house solutions, organizations are increasingly moving towards managed services for cyber defense. Security Operations Centers are specialized cybersecurity units responsible for the defense of an organization, but the large-scale centralization of threat detection is causing SOCs to endure an overwhelming amount of false positive alerts -- a phenomenon known as alert fatigue. Large collections of imprecise sensors, an inability to adapt to known false positives, evolution of the threat landscape, and inefficient use of analyst time all contribute to the alert fatigue problem. To combat these issues, we present That Escalated Quickly (TEQ), a machine learning framework that reduces alert fatigue with minimal changes to SOC workflows by predicting alert-level and incident-level actionability. On real-world data, the system is able to reduce the time it takes to respond to actionable incidents by $22.9\%$, suppress $54\%$ of false positives with a $95.1\%$ detection rate, and reduce the number of alerts an analyst needs to investigate within singular incidents by $14\%$.Open AccessThat Escalated Quickly: An ML Framework for Alert Prioritization
In place of in-house solutions, organizations are increasingly moving towards managed services for cyber defense. Security Operations Centers are specialized cybersecurity units responsible for the defense of an organization, but the large-scale centralization of threat detection is causing SOCs to endure an overwhelming amount of false positive alerts -- a phenomenon known as alert fatigue. Large collections of imprecise sensors, an inability to adapt to known false positives, evolution of the threat landscape, and inefficient use of analyst time all contribute to the alert fatigue problem. To combat these issues, we present That Escalated Quickly (TEQ), a machine learning framework that reduces alert fatigue with minimal changes to SOC workflows by predicting alert-level and incident-level actionability. On real-world data, the system is able to reduce the time it takes to respond to actionable incidents by $22.9\%$, suppress $54\%$ of false positives with a $95.1\%$ detection rate, and reduce the number of alerts an analyst needs to investigate within singular incidents by $14\%$.