Generative AI tools have increased the risk of BEC, and traditional cybersecurity defenses struggle to stay ahead of the growing speed, scale, and sophistication of attacks. Only multilayered, defense-in-depth strategies can counter the AI-powered BEC threat.
As people continue to be the weak link in most organizations’ cybersecurity practices, the growing use of generative AI tools in cyber-attacks makes email, their primary communications channel, a more compelling target than ever. The risk associated with Business Email Compromise (BEC) in particular continues to rise as generative AI tools equip attackers to build and launch social engineering and phishing campaigns with greater speed, scale, and sophistication.
What is BEC?
BEC is defined in different ways, but generally refers to cyber-attacks in which attackers abuse email — and users’ trust — to trick employees into transferring funds or divulging sensitive company data.
Unlike generic phishing emails, most BEC attacks do not rely on “spray and pray” dissemination or on users’ clicking bogus links or downloading malicious attachments. Instead, modern BEC campaigns use a technique called “pretexting.”
What is pretexting?
Pretexting is a more specific form of phishing that describes an urgent but false situation — the pretext — that requires the transfer of funds or revelation of confidential data.
This type of attack, and therefore BEC, is dominating the email threat landscape. As reported in Verizon’s 2024 Data Breach Investigation Report, recently there has been a “clear overtaking of pretexting as a more likely social action than phishing.” The data shows pretexting, “continues to be the leading cause of cybersecurity incidents (accounting for 73% of breaches)” and one of “the most successful ways of monetizing a breach.”
Pretexting and BEC work so well because they exploit humans’ natural inclination to trust the people and companies they know. AI compounds the risk by making it easier for attackers to mimic known entities and harder for security tools and teams – let alone unsuspecting recipients of routine emails – to tell the difference.
BEC attacks now incorporate AI
With the growing use of AI by threat actors, trends point to BEC gaining momentum as a threat vector and becoming harder to detect. By adding ingenuity, machine speed, and scale, generative AI tools like OpenAI’s ChatGPT give threat actors the ability to create more personalized, targeted, and convincing emails at scale.
Large Language Models (LLMs) like ChatGPT can draft believable messages that feel like emails that target recipients expect to receive. For example, generative AI tools can be used to send fake invoices from vendors known to be involved with well-publicized construction projects. These messages also prove harder to detect as AI automatically:
Avoids misspellings and grammatical errors
Creates multiple variations of email text
Translates messages that read well in multiple languages
And accomplishes additional, more targeted tactics
AI creates a force multiplier that allows primitive mass-mail campaigns to evolve into sophisticated automated attacks. Instead of spending weeks studying the target to craft an effective email, cybercriminals might only spend an hour or two and achieve a better result.
Challenges of detecting AI-powered BEC attacks
Rules-based detections miss unknown attacks
One major challenge comes from the fact that rules based on known attacks have no basis to deny new threats. While native email security tools defend against known attacks, many modern BEC attacks use entirely novel language and can omit payloads altogether. Instead, they rely on pure social engineering or bide their time until security tools recognize the new sender as a legitimate contact.
Most defensive AI can’t keep pace with attacker innovation
Security tools might focus on the meaning of an email’s text in trying to recognize a BEC attack, but defenders still end up in a rules and signature rat race. Some newer Integrated Cloud Email Security (ICES) vendors attempt to use AI defensively to improve the flawed approach of only looking for exact matches. Employing data augmentation to identify similar-looking emails helps to a point but not enough to outpace novel attacks built with generative AI.
What tools can stop BEC?
A modern defense-in-depth strategy must use AI to counter the impact of AI in the hands of attackers. As found in our 2024 State of AI Cybersecurity Report, 96% of survey participants believe AI-driven security solutions are a must have for countering AI-powered threats.
However, not all AI tools are the same. Since BEC attacks continue to change, defensive AI-powered tools should focus less on learning what attacks look like, and more on learning normal behavior for the business. By understanding expected behavior on the company’s side, the security solution will be able to recognize anomalous and therefore suspicious activity, regardless of the word choice or payload type.
To combat the speed and scale of new attacks, an AI-led BEC defense should spot novel threats.
Self-Learning AI builds profiles for every email user, including their relationships, tone and sentiment, content, and link sharing patterns. Rich context helps in understanding how people communicate and identifying deviations from the normal routine to determine what does and does not belong in an individual’s inbox and outbox.
Other email security vendors may claim to use behavioral AI and unsupervised machine learning in their products, but their AI are still pre-trained with historical data or signatures to recognize malicious activity, rather than demonstrating a true learning process. Darktrace’s Self Learning-AI truly learns from the organization in which it is installed, allowing it to detect unknown and novel vectors that other security tools are not yet trained on.
Because Darktrace understands the human behind email communications rather than knowledge of past attacks, Darktrace / EMAIL can stop the most sophisticated and evolving email security risks. It enhances your native email security by leveraging business-centric behavioral anomaly detection across inbound, outbound, and lateral messages in both email and Teams.
This unique approach quickly identifies sophisticated threats like BEC, ransomware, phishing, and supply chain attacks without duplicating existing capabilities or relying on traditional rules, signatures, and payload analysis.
The power of Darktrace’s AI can be seen in its speed and adaptability: Darktrace / EMAIL blocks the most novel threats up to 13 days faster than traditional security tools.
Oops! Something went wrong while submitting the form.
Newsletter
Stay ahead of threats with the Darktrace blog newsletter
Get the latest insights from the cybersecurity landscape, including threat trends, incident analysis, and the latest Darktrace product developments – delivered directly to your inbox, monthly.
Thanks, your request has been received
A member of our team will be in touch with you shortly.
Oops! Something went wrong while submitting the form.
Inside the SOC
Darktrace cyber analysts are world-class experts in threat intelligence, threat hunting and incident response, and provide 24/7 SOC support to thousands of Darktrace customers around the globe. Inside the SOC is exclusively authored by these experts, providing analysis of cyber incidents and threat trends, based on real-world experience in the field.
Author
Carlos Gray
Product Manager
Carlos Gonzalez Gray is a Product Marketing Manager at Darktrace, based in the Madrid Office. As an email security Subject Matter Expert he collaborates with the global product team to align each product with the company’s ethos and ensures Darktrace are continuously pushing the boundaries of innovation. His prior role at Darktrace was in Sales Engineering, leading the Iberian team and specializing in both the email and OT sectors. Additionally, his prior experience as a consultant to IBEX 35 companies in Spain has made him well-versed in compliance, auditing, and data privacy. Carlos holds an Honors BA in Political Science and a Masters in Cybersecurity from IE University.
Detecting and mitigating adversary-in-the-middle phishing attacks with Darktrace Services
What is an Adversary-in-the-Middle Attack?
Threat actors are increasingly utilizing advanced phishing toolkits and techniques to carry out Adversary-in-the-Middle (AitM) attacks. These attacks involve the use of a proxy to a legitimate service, where the attacker’s webpage mimics the expected site. While the victim believes they are visiting the legitimate site, they are actually interacting with the attacker’s device, allowing the malicious actor to monitor all interactions and control the authenticated session, ultimately gaining access to the user’s account [1][2].
In these types of AitM BEC attacks, recipients are often targeted with Dropbox-related emails, featuring subject headings like ‘FirstLast shared "Filename" with you,’ which suggest an individual is sharing an invoice-related attachment. These email subjects are common in such attacks, as threat actors attempt to encourage victims to access Dropbox links by masquerading them as legitimate files.
While higher priority users are, of course, targeted, the scope of these attacks remains broad. For instance, if a lower priority user is targeted by a phishing attack or their token is stolen, an attacker can still attempt BEC for further malicious intent and financial gain.
In October 2024, a Darktrace customer received a phishing email from a seemingly legitimate Dropbox address. This email originated from the IP, 54.240.39[.]219 and contained multiple link payloads to Dropbox-related hostnames were observed, inviting the user to access a file. Based on anomaly indicators and detection by Darktrace / EMAIL, Darktrace recognized that one of the payloads was attempting to abuse a legitimate cloud platform to share files or other unwanted material with the recipient.
Figure 1: Overview of the malicious email in the Darktrace / EMAIL console, highlighting Dropbox associated content/link payloads.
Following the recipient’s engagement with this email, Darktrace / IDENTITY identified a series of suspicious activities within the customer’s environment.
AitM attacks allow threat actors to bypass multi-factor authentication (MFA). Initially, when a user is phished, the malicious infrastructure captures both the user’s credentials and the token. This includes replaying a token issued to user that has already completed the MFA requirement, allowing the threat actor to satisfy the validity of the requirement and gain access to sensitive organizational resources. Darktrace is able to analyze user activity and authentication patterns to determine whether MFA requirements were met. This capability helps verify and indicate token theft via AitM.
Darktrace observed the associated user account making requests over Microsoft 365 from the IP 41.90.175[.]46. Given the unusual nature and rare geolocation based in Kenya, Africa, this activity did not appear indicative of legitimate business operations.
Figure 2: Geographical location of the SaaS user in relation to the source IP 41.90.175[.]46.
Further analysis using open-source intelligence (OSINT) revealed that the endpoint was likely associated with a call-back proxy network [3]. This suggested the presence of a network device capable of re-routing traffic and harvesting information.
Darktrace also detected that the same SaaS user was logging in from two different locations around the same time. One login was from a common, expected location, while the other was from an unusual location. Additionally, the user was observed registering security information using the Microsoft Authenticator app, indicating an attempt by an attacker to maintain access to the account by establishing a new method of MFA. This new MFA method could be used to bypass future MFA requirements, allowing the attacker to access sensitive material or carry out further malicious activities.
Figure 3: External sites summary for the SaaS account in relation to the source IP 13.74.161[.]104, observed with Registering Security Information.
Ultimately, this anomalous behavior was escalated to the Darktrace Security Operations Centre (SOC) via the Managed Detection & Response service for prompt triage and investigation by Darktrace’s SOC Analysts who notified the customer of strong evidence of compromise.
Fortunately, since this customer had Darktrace enabled in Autonomous Response mode, the compromised SaaS account had already been disabled, containing the attack. Darktrace’s SOC elected to extend this action to ensure the malicious activity remained halted until the customer could take further remedial action.
Figure 4: Attack timeline of observed activity, in chronological order; This highlighted anomalous SaaS events such as, MailItemsAccessed’, ‘Use of Unusual Credentials’, ‘User Registered Security Info’ events, and a ‘Disable User’ Autonomous Response action.
Conclusion
AitM attacks can play a crucial role in BEC campaigns. These attacks are often part of multi-staged operations, where an initial AitM attack is leveraged to launch a BEC by delivering a malicious URL through a trusted vendor or service. Attackers often attempt to lay low on their target network, sometimes persisting for extended periods, as they monitor user accounts or network segments to intercept sensitive communications.
In this instance, Darktrace successfully identified and acted against AitM techniques being leveraged in a BEC attack that used Dropbox for delivery. While Dropbox is widely used for legitimate purposes, its popularity has also made it a target for exploitation by threat actors, who have used it for a variety of malicious purposes, including delivering malware and revealing sensitive information.
Darktrace’s Security Operations Support service, combined with its Autonomous Response technology, provided timely and effective mitigation. Dedicated Security Operations Support analysts triaged the incident and implemented preventative measures, ensuring the customer was promptly notified. Meanwhile, Darktrace swiftly disabled the compromised SaaS account, allowing the customer to take further necessary actions, such as resetting the user’s password.
This case highlights the capabilities of Darktrace’s solutions, enabling the customer to resume normal business operations despite the malicious activity.
Credit to Justin Torres (Senior Cyber Analyst), Stefan Rowe (Technical Director, SOC) and Ryan Traill (Analyst Content Lead)
A Snake in the Net: Defending Against AiTM Phishing Threats and Mamba 2FA
What are Adversary-in-the-Middle (AiTM) phishing kits?
Phishing-as-a-Service (PhaaS) platforms have significantly lowered the barriers to entry for cybercriminals, enabling a new wave of sophisticated phishing attacks. Among the most concerning developments in this landscape is the emergence of Adversary-in-the-Middle (AiTM) phishing kits, which enhance traditional phishing tactics by allowing attackers to intercept and manipulate communications in real-time. The PhaaS marketplace offers a wide variety of innovative capabilities, with basic services starting around USD 120 and more advanced services costing around USD 250 monthly [1].
These AiTM kits are designed to create convincing decoy pages that mimic legitimate login interfaces, often pre-filling user information to increase credibility. By acting as a man-in-the-middle, attackers can harvest sensitive data such as usernames, passwords, and even multi-factor authentication (MFA) tokens without raising immediate suspicion. This capability not only makes AiTM attacks more effective but also poses a significant challenge for cybersecurity defenses [2].
Mamba 2FA is one such example of a PhaaS strain with AiTM capabilities that has emerged as a significant threat to users of Microsoft 365 and other enterprise systems. Discovered in May 2024, Mamba 2FA employs advanced AiTM tactics to bypass MFA, making it particularly dangerous for organizations relying on these security measures.
What is Mamba 2FA?
Phishing Mechanism
Mamba 2FA employs highly convincing phishing pages that closely mimic legitimate Microsoft services like OneDrive and SharePoint. These phishing URLs are crafted with a specific structure, incorporating Base64-encoded parameters. This technique allows attackers to tailor the phishing experience to the targeted organization, making the deception more effective. If an invalid parameter is detected, users are redirected to a benign error page, which helps evade automated detection systems [5].
Figure 1: Phishing page mimicking the Microsoft OneDrive service.
Real-Time Communication
A standout feature of Mamba 2FA is its use of the Socket.IO JavaScript library. This library facilitates real-time communication between the phishing page and the attackers' backend servers. As users input sensitive information, such as usernames, passwords, and MFA tokens on the phishing site, this data is immediately relayed to the attackers, enabling swift unauthorized access [5].
Multi-Factor Authentication Bypass
Mamba 2FA specifically targets MFA methods that are not resistant to phishing, such as one-time passwords (OTPs) and push notifications. When a user enters their MFA token, it is captured in real-time by the attackers, who can then use it to access the victim's account immediately. This capability significantly undermines traditional security measures that rely on MFA for account protection.
Infrastructure and Distribution
The platform's infrastructure consists of two main components: link domains and relay servers. Link domains handle initial phishing attempts, while relay servers are responsible for stealing credentials and completing login processes on behalf of the attacker. The relay servers are designed to mask their IP addresses by using proxy services, making it more difficult for security systems to block them [3].
Evasion Techniques
To evade detection by security tools, Mamba 2FA employs several strategies:
Sandbox Detection: The platform can detect if it is being analyzed in a sandbox environment and will redirect users to harmless pages like Google’s 404 error page.
Dynamic URL Generation: The URLs used in phishing attempts are frequently rotated and often short-lived to avoid being blacklisted by security solutions.
HTML Attachments: Phishing emails often include HTML attachments that appear benign but contain hidden JavaScript that redirects users to the phishing page [5].
Darktrace’s Coverage of Mamba 2FA
Starting in July 2024, the Darktrace Threat Research team detected a sudden rise in Microsoft 365 customer accounts logging in from unusual external sources. These accounts were accessed from an anomalous endpoint, 2607:5500:3000:fea[::]2, and exhibited unusual behaviors upon logging into Software-as-a-Service (SaaS) accounts. This activity strongly correlates with a phishing campaign using Mamba 2FA, first documented in late June 2024 and tracked as Mamba 2FA by Sekoia [2][3].
Darktrace / IDENTITY was able to identify the initial stages of the Mamba 2FA campaign by correlating subtle anomalies, such as unusual SaaS login locations. Using AI based on peer group analysis, it detected unusual behavior associated with these attacks. By leveraging Autonomous Response actions, Darktrace was able to neutralize these threats in every instance of the campaign detected.
On July 23, a SaaS user was observed logging in from a rare ASN and IP address, 2607:5500:3000:fea::2, originating from the US and successfully passed through MFA authentication.
Figure 2: Model Alert Event Log showing Darktrace’s detection of a SaaS user mailbox logging in from an unusual source it correlates with Mamba 2FA relay server.
Almost an hour later, the SaaS user was observed logging in from another suspicious IP address, 45.133.172[.]86, linked to ASN AS174 COGENT-174. This IP, originating from the UK, successfully passed through MFA validation.
Following this unusual access, the SaaS user was notably observed reading emails and files that could contain sensitive payment and contract information. This behavior suggests that the attacker may have been leveraging contextual information about the target to craft further malicious phishing emails or fraudulent invoices. Subsequently, the user was detected creating a new mailbox rule titled 'fdsdf'. This rule was configured to redirect emails from a specific domain to the 'Deleted Items' folder and automatically mark them as read.
Implications of Unusual Email Rules
Such unusual email rule configurations are a common tactic employed by attackers. They often use these rules to automatically forward emails containing sensitive keywords—such as "invoice”, "payment", or "confidential"—to an external address. Additionally, these rules help conceal malicious activities, keeping them hidden from the target and allowing the attacker to operate undetected.
Figure 3: The model alert “SaaS / Compliance / Anomalous New Email Rule,” pertaining to the unusual email rule created by the SaaS user named ‘fdsdf’.
Blocking the action
A few minutes later, the SaaS user from the unusual IP address 45.133.172[.]86 was observed attempting to send an email with the subject “RE: Payments.” Subsequently, Darktrace detected the user engaging in activities that could potentially establish persistence in the compromised account, such as registering a new authenticator app. Recognizing this sequence of anomalous behaviors, Darktrace implemented an Autonomous Response inhibitor, disabling the SaaS user for two hours. This action effectively contained potential malicious activities, such as the distribution of phishing emails and fraudulent invoices, and gave the customer’s security team the necessary time to conduct a thorough investigation and implement appropriate security measures.
Figure 4: Device Event Log displaying Darktrace’s Autonomous Response taking action by blocking the SaaS account.
Figure 5: Darktrace / IDENTITY highlighting the 16 model alerts that triggered during the observed compromise.
In another example from mid-July, similar activities related to the campaign were observed on another customer network. A SaaS user was initially detected logging in from the unusual external endpoint 2607:5500:3000:fea[::]2.
Figure 6: The SaaS / Compromise / SaaS Anomaly Following Anomalous Login model alert was triggered by an unusual login from a suspicious IP address linked to Mamba 2FA.
A few minutes later, in the same manner as demonstrated in the previous case, the actor was observed logging in from another rare endpoint, 102.68.111[.]240. However, this time it was from a source IP located in Lagos, Nigeria, which no other user on the network had been observed connecting from. Once logged in, the SaaS user updated the settings to "User registered Authenticator App with Notification and Code," a possible attempt to maintain persistence in the SaaS account.
Figure 7: Darktrace / IDENTITY highlighted the regular locations for the SaaS user. The rarity scores associated with the Mamba 2FA IP location and another IP located in Nigeria were classified as having very low regularity scores for this user.
Based on unusual patterns of user behavior, a Cyber AI Analyst Incident was also generated, detailing all potential account hijacking activities. Darktrace also applied an Autonomous Response action, disabling the user for over five hours. This swift action was crucial in preventing further unauthorized access, potential data breaches and further implications.
Figure 8: Cyber AI Analyst Incident detailing the unusual activities related to the SaaS account hijacking.
Since the customer had subscribed to Darktrace Security Operations Centre (SOC) services, Darktrace analysts conducted an additional human investigation confirming the account compromise.
How Darktrace Combats Phishing Threats
The initial entry point for Mamba 2FA account compromises primarily involves phishing campaigns using HTML attachments and deceptive links. These phishing attempts are designed to mimic legitimate Microsoft services, such as OneDrive and SharePoint, making them appear authentic to unsuspecting users. Darktrace / EMAIL leverages multiple capabilities to analyze email content for known indicators of phishing. This includes looking for suspicious URLs, unusual attachments (like HTML files with embedded JavaScript), and signs of social engineering tactics commonly used in phishing campaigns like Mamba 2FA. With these capabilities, Darktrace successfully detected Mamba 2FA phishing emails in networks where this tool is integrated into the security layers, consequently preventing further implications and account hijacks of their users.
Mamba 2FA URL Structure and Domain Names
The URL structure used in Mamba 2FA phishing attempts is specifically designed to facilitate the capture of user credentials and MFA tokens while evading detection. These phishing URLs typically follow a pattern that incorporates Base64-encoded parameters, which play a crucial role in the operation of the phishing kit.
The URLs associated with Mamba 2FA phishing pages generally follow this structure [6]:
https://{domain}/{m,n,o}/?{Base64 string}
Below are some potential Mamba 2FA phishing emails, with the Base64 strings already decoded, that were classified as certain threats by Darktrace / EMAIL. This classification was based on identifying multiple suspicious characteristics, such as HTML attachments containing JavaScript code, emails from senders with no previous association with the recipients, analysis of redirect links, among others. These emails were autonomously blocked from being delivered to users' inboxes.
Figure 9: Darktrace / EMAIL highlighted a possible phishing email from Mamba 2FA, which was classified as a 100% anomaly.
Figure 10: Darktrace / EMAIL highlighted a URL that resembles the characteristics associated with Mamba 2FA.
Conclusion
The rise of PhaaS platforms and the advent of AiTM phishing kits represent a concerning evolution in cyber threats, pushing the boundaries of traditional phishing tactics and exposing significant vulnerabilities in current cybersecurity defenses. The ability of these attacks to effortlessly bypass traditional security measures like MFA underscores the need for more sophisticated, adaptive strategies to combat these evolving threats.
By identifying and responding to anomalous activities within Microsoft 365 accounts, Darktrace not only highlights the importance of comprehensive monitoring but also sets a new standard for proactive threat detection. Furthermore, the autonomous threat response capabilities and the exceptional proficiency of Darktrace / EMAIL in intercepting and neutralizing sophisticated phishing attacks illustrate a robust defense mechanism that can effectively safeguard users and maintain the integrity of digital ecosystems.
Credit to Patrick Anjos (Senior Cyber Analyst) and Nahisha Nobregas (Senior Cyber Analyst)
Appendices
Darktrace Model Detections
SaaS / Access / M365 High Risk Level Login
SaaS / Access / Unusual External Source for SaaS Credential Use
SaaS / Compromise / Login From Rare Endpoint While User Is Active
SaaS / Compliance / M365 Security Information Modified
SaaS / Compromise / Unusual Login and New Email Rule
