The increasing capabilities of LLMs have led to the rapid proliferation of LLM agent apps, where developers enhance LLMs with access to external resources to support complex task execution. Among these, LLM email agent apps represent one of the widely used categories, as email remains a critical communication medium for users. LLM email agents are capable of managing and responding to email using LLM-driven reasoning and autonomously executing user instructions via external email APIs (e.g., send email). However, despite their growing deployment and utility, the security mechanism of LLM email agent apps remains underexplored. Currently, there is no comprehensive study into the potential security risk within these agent apps and their broader implications. In this paper, we conduct the first in-depth and systematic security study of LLM email agents. We propose the Email Agent Hijacking (EAH) attack, which overrides the original prompts of the email agent via external email resources, allowing attackers to gain control of the email agent remotely and further perform specific attack scenarios without user awareness. To facilitate the large-scale evaluation, we propose EAHawk, a pipe
Phishing email attacks are among the most common and most harmful cybersecurity attacks. With the emergence of generative AI, phishing attacks can be based on emails generated automatically, making it more difficult to detect them. That is, instead of a single email format sent to a large number of recipients, generative AI can be used to send each potential victim a different email, making it more difficult for cybersecurity systems to identify the scam email before it reaches the recipient. Here we describe a corpus of AI-generated phishing emails. We also use different machine learning tools to test the ability of automatic text analysis to identify AI-generated phishing emails. The results are encouraging, and show that machine learning tools can identify an AI-generated phishing email with high accuracy compared to regular emails or human-generated scam email. By applying descriptive analytic, the specific differences between AI-generated emails and manually crafted scam emails are profiled, and show that AI-generated emails are different in their style from human-generated phishing email scams. Therefore, automatic identification tools can be used as a warning for the user. T
Phishing attacks on enterprise employees present one of the most costly and potent threats to organizations. We explore an understudied facet of enterprise phishing attacks: the email relay infrastructure behind successfully delivered phishing emails. We draw on a dataset spanning one year across thousands of enterprises, billions of emails, and over 800,000 delivered phishing attacks. Our work sheds light on the network origins of phishing emails received by real-world enterprises, differences in email traffic we observe from networks sending phishing emails, and how these characteristics change over time. Surprisingly, we find that over one-third of the phishing email in our dataset originates from highly reputable networks, including Amazon and Microsoft. Their total volume of phishing email is consistently high across multiple months in our dataset, even though the overwhelming majority of email sent by these networks is benign. In contrast, we observe that a large portion of phishing emails originate from networks where the vast majority of emails they send are phishing, but their email traffic is not consistent over time. Taken together, our results explain why no singular de
Despite technical and non-technical countermeasures, humans continue to be tricked by phishing emails. How users make email response decisions is a missing piece in the puzzle to identifying why people still fall for phishing emails. We conducted an empirical study using a think-aloud method to investigate how people make 'response decisions' while reading emails. The grounded theory analysis of the in-depth qualitative data has enabled us to identify different elements of email users' decision-making that influence their email response decisions. Furthermore, we developed a theoretical model that explains how people could be driven to respond to emails based on the identified elements of users' email decision-making processes and the relationships uncovered from the data. The findings provide deeper insights into phishing email susceptibility due to people's email response decision-making behavior. We also discuss the implications of our findings for designers and researchers working in anti-phishing training, education, and awareness interventions
The proliferation of phishing sites and emails poses significant challenges to existing cybersecurity efforts. Despite advances in malicious email filters and email security protocols, problems with oversight and false positives persist. Users often struggle to understand why emails are flagged as potentially fraudulent, risking the possibility of missing important communications or mistakenly trusting deceptive phishing emails. This study introduces ChatSpamDetector, a system that uses large language models (LLMs) to detect phishing emails. By converting email data into a prompt suitable for LLM analysis, the system provides a highly accurate determination of whether an email is phishing or not. Importantly, it offers detailed reasoning for its phishing determinations, assisting users in making informed decisions about how to handle suspicious emails. We conducted an evaluation using a comprehensive phishing email dataset and compared our system to several LLMs and baseline systems. We confirmed that our system using GPT-4 has superior detection capabilities with an accuracy of 99.70%. Advanced contextual interpretation by LLMs enables the identification of various phishing tactic
The email system is the central battleground against phishing and social engineering attacks, and yet email providers still face key challenges to authenticate incoming emails. As a result, attackers can apply spoofing techniques to impersonate a trusted entity to conduct highly deceptive phishing attacks. In this work, we study email spoofing to answer three key questions: (1) How do email providers detect and handle forged emails? (2) Under what conditions can forged emails penetrate the defense to reach user inbox? (3) Once the forged email gets in, how email providers warn users? Is the warning truly effective? We answer these questions through end-to-end measurements on 35 popular email providers (used by billions of users), and extensive user studies (N = 913) that consist of both simulated and real-world phishing experiments. We have four key findings. First, most popular email providers have the necessary protocols to detect spoofing, but still allow forged emails to get into user inbox (e.g., Yahoo Mail, iCloud, Gmail). Second, once a forged email gets in, most email providers have no warnings for users, particularly on mobile email apps. Some providers (e.g., Gmail Inbox)
The never-ending barrage of malicious emails, such as spam and phishing, is of constant concern for users, who rely on countermeasures such as email filters to keep the intended recipient safe. Modern email filters, one of our few defence mechanisms against malicious emails, are often circumvented by sophisticated attackers. This study focuses on how attackers exploit HTML and CSS in emails to conceal arbitrary content, allowing for multiple permutations of a malicious email, some of which may evade detection by email filters. This concealed content remains undetected by the recipient, presenting a serious security risk. Our research involved developing and applying an email sampling and analysis procedure to a large-scale dataset of unsolicited emails. We then identify the sub-types of concealment attackers use to conceal content and the HTML and CSS tricks employed.
Email phishing is one of the most prevalent and globally consequential vectors of cyber intrusion. As systems increasingly deploy Large Language Models (LLMs) applications, these systems face evolving phishing email threats that exploit their fundamental architectures. Current LLMs require substantial hardening before deployment in email security systems, particularly against coordinated multi-vector attacks that exploit architectural vulnerabilities. This paper proposes LLMPEA, an LLM-based framework to detect phishing email attacks across multiple attack vectors, including prompt injection, text refinement, and multilingual attacks. We evaluate three frontier LLMs (e.g., GPT-4o, Claude Sonnet 4, and Grok-3) and comprehensive prompting design to assess their feasibility, robustness, and limitations against phishing email attacks. Our empirical analysis reveals that LLMs can detect the phishing email over 90% accuracy while we also highlight that LLM-based phishing email detection systems could be exploited by adversarial attack, prompt injection, and multilingual attacks. Our findings provide critical insights for LLM-based phishing detection in real-world settings where attackers
In the pursuit of an effective spam detection system, the focus has often been on identifying known spam patterns either through rule-based detection systems or machine learning (ML) solutions that rely on keywords. However, both systems are susceptible to evasion techniques and zero-day attacks that can be achieved at low cost. Therefore, an email that bypassed the defense system once can do it again in the following days, even though rules are updated or the ML models are retrained. The recurrence of failures to detect emails that exhibit layout similarities to previously undetected spam is concerning for customers and can erode their trust in a company. Our observations show that threat actors reuse email kits extensively and can bypass detection with little effort, for example, by making changes to the content of emails. In this work, we propose an email visual similarity detection approach, named Pisco, to improve the detection capabilities of an email threat defense system. We apply our proof of concept to some real-world samples received from different sources. Our results show that email kits are being reused extensively and visually similar emails are sent to our customers
Phishing is one of the main points of compromise, with email security and awareness being estimated at \$50-100B in 2022. There is great need for email forensics capability to quickly search for malicious content. A novel solution POST is proposed. POST is an API driven serverless email archival, processing, and flagging workflow for both large and small organizations that collects and parses all email, flags emails using state of the art Natural Language Processing and Machine Learning, allows full email searching on every aspect of an email, and provides a cost savings of up to 68.6%.
The critical role played by email has led to a range of extension protocols (e.g., SPF, DKIM, DMARC) designed to protect against the spoofing of email sender domains. These protocols are complex as is, but are further complicated by automated email forwarding -- used by individual users to manage multiple accounts and by mailing lists to redistribute messages. In this paper, we explore how such email forwarding and its implementations can break the implicit assumptions in widely deployed anti-spoofing protocols. Using large-scale empirical measurements of 20 email forwarding services (16 leading email providers and four popular mailing list services), we identify a range of security issues rooted in forwarding behavior and show how they can be combined to reliably evade existing anti-spoofing controls. We further show how these issues allow attackers to not only deliver spoofed email messages to prominent email providers (e.g., Gmail, Microsoft Outlook, and Zoho), but also reliably spoof email on behalf of tens of thousands of popular domains including sensitive domains used by organizations in government (e.g., state.gov), finance (e.g., transunion.com), law (e.g., perkinscoie.com
Email is a widely used tool for business communication, and email marketing has emerged as a cost-effective strategy for enterprises. While previous studies have examined factors affecting email marketing performance, limited research has focused on understanding email response behavior by considering email content and metadata. This study proposes a Prototype-based Multi-view Network (PROMINET) that incorporates semantic and structural information from email data. By utilizing prototype learning, the PROMINET model generates latent exemplars, enabling interpretable email response prediction. The model maps learned semantic and structural exemplars to observed samples in the training data at different levels of granularity, such as document, sentence, or phrase. The approach is evaluated on two real-world email datasets: the Enron corpus and an in-house Email Marketing corpus. Experimental results demonstrate that the PROMINET model outperforms baseline models, achieving a ~3% improvement in F1 score on both datasets. Additionally, the model provides interpretability through prototypes at different granularity levels while maintaining comparable performance to non-interpretable mod
Email platforms need to generate personalized rankings of emails that satisfy user preferences, which may vary over time. We approach this as a recommendation problem based on three criteria: closeness (how relevant the sender and topic are to the user), timeliness (how recent the email is), and conciseness (how brief the email is). We propose MOSR (Multi-Objective Stationary Recommender), a novel online algorithm that uses an adaptive control model to dynamically balance these criteria and adapt to preference changes. We evaluate MOSR on the Enron Email Dataset, a large collection of real emails, and compare it with other baselines. The results show that MOSR achieves better performance, especially under non-stationary preferences, where users value different criteria more or less over time. We also test MOSR's robustness on a smaller down-sampled dataset that exhibits high variance in email characteristics, and show that it maintains stable rankings across different samples. Our work offers novel insights into how to design email re-ranking systems that account for multiple objectives impacting user satisfaction.
Email phishing has become more prevalent and grows more sophisticated over time. To combat this rise, many machine learning (ML) algorithms for detecting phishing emails have been developed. However, due to the limited email data sets on which these algorithms train, they are not adept at recognising varied attacks and, thus, suffer from concept drift; attackers can introduce small changes in the statistical characteristics of their emails or websites to successfully bypass detection. Over time, a gap develops between the reported accuracy from literature and the algorithm's actual effectiveness in the real world. This realises itself in frequent false positive and false negative classifications. To this end, we propose a multidimensional risk assessment of emails to reduce the feasibility of an attacker adapting their email and avoiding detection. This horizontal approach to email phishing detection profiles an incoming email on its main features. We develop a risk assessment framework that includes three models which analyse an email's (1) threat level, (2) cognitive manipulation, and (3) email type, which we combine to return the final risk assessment score. The Profiler does no
In this paper, the process of converting the Enron email dataset (the version cited in the preprint) to thousands of features per email for a selected set of 2400 labelled emails is explained and evaluated. The final features are tailored for Cosine distance so that the Cosine distance invertly reflect the number of top indicative words of each email that are common between the two emails in an explainable normalized fashion. The labelling is based on the leaf folder name in the Enron email dataset (the version cited in the preprint) folders tree and the 2400 emails selected consist 300 emails for each of the 8 labels. The evaluation is based on the accuracy of a k nearest neighbours majority voting classification using Cosine distance. In addition to KNN majority voting classification accuracy and confusion matrix, some statistics for the process is reported. The KNN majority voting classification accuracy using Cosine distance is 76.75% which shows at least some level of success given the 8 labels involved. The result of conversion is 48557 features per selected email out of which exactly 40 features per email are non-zero. The result of conversion is a data set named MeeefTCD (M
Anomalies in emails such as phishing and spam present major security risks such as the loss of privacy, money, and brand reputation to both individuals and organizations. Previous studies on email anomaly detection relied on a single type of anomaly and the analysis of the email body and subject content. A drawback of this approach is that it takes into account the written language of the email content. To overcome this deficit, this study conducted feature extraction and selection on email header datasets and leveraged both multi and one-class anomaly detection approaches. Experimental analysis results obtained demonstrate that email header information only is enough to reliably detect spam and phishing emails. Supervised learning algorithms such as Random Forest, SVM, MLP, KNN, and their stacked ensembles were found to be very successful, achieving high accuracy scores of 97% for phishing and 99% for spam emails. One-class classification with One-Class SVM achieved accuracy scores of 87% and 89% with spam and phishing emails, respectively. Real-world email filtering applications will benefit from the use of only the header information in terms of resources utilization and efficie
Emails are used every day for communication, and many countries and organisations mostly use email for official communications. It is highly valued and recognised for confidential conversations and transactions in day-to-day business. The Often use of this channel and the quality of information it carries attracted cyber attackers to it. There are many existing techniques to mitigate attacks on email, however, the systems are more focused on email content and behaviour and not securing entrances to email boxes, composition, and settings. This work intends to protect users' email composition and settings to prevent attackers from using an account when it gets hacked or hijacked and stop them from setting forwarding on the victim's email account to a different account which automatically stops the user from receiving emails. A secure code is applied to the composition send button to curtail insider impersonation attack. Also, to secure open applications on public and private devices.
One of the ways in which attackers try to steal sensitive information from corporations is by sending spearphishing emails. This type of emails typically appear to be sent by one of the victim's coworkers, but have instead been crafted by an attacker. A particularly insidious type of spearphishing emails are the ones that do not only claim to come from a trusted party, but were actually sent from that party's legitimate email account that was compromised in the first place. In this paper, we propose a radical change of focus in the techniques used for detecting such malicious emails: instead of looking for particular features that are indicative of attack emails, we look for possible indicators of impersonation of the legitimate owners. We present IdentityMailer, a system that validates the authorship of emails by learning the typical email-sending behavior of users over time, and comparing any subsequent email sent from their accounts against this model. Our experiments on real world e-mail datasets demonstrate that our system can effectively block advanced email attacks sent from genuine email accounts, which traditional protection systems are unable to detect. Moreover, we show
In this work, we present the first dataset, MailEx, for performing event extraction from conversational email threads. To this end, we first proposed a new taxonomy covering 10 event types and 76 arguments in the email domain. Our final dataset includes 1.5K email threads and ~4K emails, which are annotated with totally ~8K event instances. To understand the task challenges, we conducted a series of experiments comparing three types of approaches, i.e., fine-tuned sequence labeling, fine-tuned generative extraction, and few-shot in-context learning. Our results showed that the task of email event extraction is far from being addressed, due to challenges lying in, e.g., extracting non-continuous, shared trigger spans, extracting non-named entity arguments, and modeling the email conversational history. Our work thus suggests more future investigations in this domain-specific event extraction task.
Email is one of the most successful computer applications yet devised. Communication features in email, however, have remained relatively static in years. We investigate one way of expanding email functionality without modifying the existing email infrastructure. We introduce email late bound content, a simple and generalizable technique that defers message content binding through image lazy-loading. Parts of an email are converted into external images embedded in HTML code snippets, making it so that email clients will defer the image download (i.e. content binding) until the moment users open the email. This late bound content allows email senders and third party services to update delivered emails. To illustrate the utilities of late bound content, we present four new example features and discuss the tradeoffs of email content late binding.