Executive Summary
Modern cybercrime has evolved from isolated hacking into a scalable, commercialized industry.
By adopting corporate business models and leveraging unregulated AI, criminal networks are bypassing traditional security controls and exposing financial institutions to unprecedented operational and reputational risk.
As part of .Bank’s ongoing commitment to strengthening trust and security across the financial sector, this article examines the commercial mechanics that are driving modern fraud. Defending against this industrialized threat requires shifting from reactive security to proactive, multi-layered resilience.
The key is to combine technical tools with targeted human awareness to protect banking assets, executives, and customers.
Introduction
The World Economic Forum’s Global Cybersecurity Outlook 2026 report identified fraud, ransomware attacks, and phishing in its several forms (including voice phishing, aka vishing, and text phishing, aka smishing) as the top cyber risk concerns for CEOs in 2026.
77% of CEOs interviewed think that the risk of cyber-enabled fraud and phishing have increased, and 62% know someone in their professional or personal networks who has been affected by phishing, vishing, or smishing attacks.
Nearly one in five CEOs knows a victim of investment scams or romance baiting. Contributors to the project concluded that “cyber-enabled fraud poses a pervasive societal threat, undermining trust and security across all demographics—from corporate leaders to households and vulnerable populations.”
Cybercriminals have evolved fraud from individuals or small groups of conspirators into a diverse, global industry; simply put, criminals have “gone commercial”. They have adopted corporate business practices and commercialized service models, operating from illicit networks that span the globe. They’ve made extensive use of Artificial Intelligence (AI) to streamline their operations.
Their path to success is going corporate.
Fraud affects victims and Financial Institutions
Source: US FBI, November 2025
Source: UK Finance Ministry, July 2026
Source: TransUnion UK, H1 2026
Cybercrime-as-a-Service (CaaS)
Criminal enterprises operating worldwide from China, Russia, North Korea, the Middle East, and Africa offer subscriber or affiliate services to other cybercriminals for phishing or ransomware attacks. These service operations emulate legitimate, commercial cloud offerings. They provide domain registration, fake website hosting with administration tools, reverse proxy for MFA bypass, and phishing kits (see sidebar). Operational AI is present on many CaaS platforms.
CaaS operations offer ready-to-roll attack campaigns. Subscribers can host phishing websites, reverse proxies, generate phishing URLs, and send email or text message “lures” that include the URL of the phishing page through the CaaS administrative (control) panel. The services are promoted on dark web forums or secure messaging channels. CaaS training is offered in Discord communities.
Subscribers pay monthly subscription or per-use fees to conduct attacks. CaaS operators pay commissions or a percentage of illicit proceeds in cryptocurrency to subscribers for successful attacks or stolen credentials.
Criminal enterprises often use guarantee services to facilitate and launder payments. One notable player, Xinbi Guarantee, is believed to have processed nearly $18 billion in crypto transactions for cybercriminal enterprises through wallet services offered via secure messaging services (Telegram and SafeW).
CaaS makes the cybercrime business highly accessible, scalable, and profitable for its operators.
What Are Attack Kits?
Attack kits are typically sets of files and scripts that provide a criminal with tools to conduct an attack quickly and easily and are usually specific to certain types of crime.
Exploit kits are malicious software that can be attached to an email or downloaded via a malicious hyperlink. This malware exploits a vulnerability in the victim’s device and installs additional malware payloads. For example, a banking information-stealing trojan or a remote administration trojan (RAT).
Phishing kits are sets of files—web pages, forms, and scripts—that are used to create phishing web sites. They may include a reverse-proxy that will be used to bypass multi-factor authentication (MFA bypass). Attack kits can be obtained in dark markets, or they may be offered as part of a crime-as-a-service.
Cybercriminals have effectively employed interception-and-relay attack techniques to steal credentials. For example, adversary-in-the-middle (AiTM) attacks use a URL embedded in a phishing email or text message to direct a user to a fake site, where the user enters credentials into what appears to be a legitimate login form. But at this fake page, a reverse proxy captures (intercepts) and relays the credentials to the legitimate site and login. The proxy performs the steps necessary to complete a multi-factor authentication with the legitimate login page to obtain a session cookie or access token and typically returns a credible post-login page. The attacker now possesses the credentials and can exploit the now authenticated session for data exfiltration or other malicious activities—for as long as the cookie or token remains valid.
The EvilProxy, TycoonMFA, and Sneaky2FA are CaaS operations known to have used AiTM. Other operations, such as Kali 365 and EvilTokens, used a different intercept-and-relay attack. Device-code phishing has been used in business email compromise (BEC) and attacks against corporations in the finance sector to steal credentials from finance or accounting officers and employees by impersonating trusted identity management platforms like Okta and Microsoft Entra ID and OAuth. Like AiTM attacks, successful device-code phishing attacks provide criminals with authenticated sessions that they can abuse for data theft, extortion, or unauthorized data disclosure.
Cryptocurrency investment (e.g., romance baiting) scams are a different form of CaaS. Here, the objective is not credential theft but tricking victims into making fraudulent funds transfers. The proxies in these scams are forced labor (often trafficked) who cultivate fake romantic or social relationships with victims and eventually persuade their victims to invest money in a fraudulent cryptocurrency or other investment scheme.
AI Improves Your Workflows… and It’s Improving Criminal Workflows as well
Corporations are adopting AI to automate workflows and to increase operational efficiency. The beneficial uses of AI are evident. For instance, in healthcare, AI can improve medical image quality and image analysis. AI has proven useful in network diagnostics, software development, and vulnerability assessment. In the financial industry, AI’s ability to quickly process large numbers of transactions improves fraud detection.
AI is also improving workflows in the criminal underworld. An Anthropic study on the misuse of AI explains how the same benefits from automation, optimization, and scale that AI offers to legitimate industries are also beneficial to organized cybercrime. However, criminals have a distinct advantage over legitimate AI users: they are unencumbered by governance frameworks, guardrails, or ethics. With virtually no data-collection prohibitions, scammers and phishers have used AI to improve attacks and operations in several ways.
- Phishing and spam messages looks authentic. Criminals use AI to correct poor spelling, grammar, cadence, and other telltale signs once common in phishing lures.
- BEC is more personalized and convincing. Phishers train their Large Language Models (LLMs) with public social media profiles and corporate web sites to generate messages that are not only highly convincing but tailored to specific targets (CEOs and CFOs, a.k.a., whale phishing).
- Deepfake audio and video vishing are indistinguishable from authentic audio and video. Criminals use generative AI to clone the voices or visual likenesses of executives, key personnel, or even interview candidates. Traditionally, humans have relied upon verbal and visual recognition to confirm who they are communicating with, but human perceptions can now be faked by AI tools. These deepfakes can trick employees into authorizing fraudulent wire transfers, or extorting companies or families.
- Ransomware groups use AI to evaluate and weaponize stolen data. LLMs efficiently identify intellectual property, financial records, or extortion-worthy personal data. Here, AI is an operational accelerator.
- AI enhances synthetic identity theft. LLMs combine fabricated photos and personas with stolen personal data to create seemingly authentic identities. Criminals can then use these synthetic identities to open accounts with banks or crypto exchanges, where the AI can take on the role of a “money mule” to launder illicit funds.
AI is already well-entrenched in spam and phishing attacks. Analysis by university and security researchers found that over 50% of spam “appeared to have been generated by a large language model” while “14% of BEC attacks were generated by AI.”
Legitimate businesses make use of AI to measure and evaluate their operations. Criminals do this as well. Attackers can use the same AI algorithms to assess outcomes—i.e., how well an attack has performed. Today, criminals that operate traffic distribution networks refine their attack strategies through conversion filtering and rule-based routing.
It’s only a matter of time before criminals run AI models on performance data to make their attacks more efficient and profitable.
Criminals also use AI tools to manage stolen funds transferred across hundreds of microtransactions, thereby optimizing transaction flows so that the transfers blend in with legitimate traffic, evading automated bank fraud detection systems.
Legitimizing Fraudulently Obtained Gains
Criminals in the fraud space prefer to be paid by victims as directly as possible, in a way that makes these payments difficult to track.
Cryptocurrency is the “coin of the realm” for global criminal organizations, who must convert illicit proceeds into cleansed financial assets, property, merchandise, or government-issued currency that they can use in the real world. To cash out (and often, to pay subscribers or affiliates), they must launder their illicit proceeds.
Criminals want laundering to appear commonplace and legitimate. They want it to go unnoticed, and they want to avoid attracting attention from regulatory authorities or law enforcement.
The cashing-out process begins with collections from victims and ends in fiat currency. Checks, credit cards, wire transfers, and physical cash are attractive, but the legacy of financial institutions and the fraud detection built into them presents challenges with these forms of currency. In practice, criminal transactions mostly avoid them.
Criminals commonly coerce ransomware victims to pay in cryptocurrency. Enforcement barriers require further extraction and transfer beyond the reach of financial institutions and law enforcement, but these pipelines are harder to detect, prevent, and reverse.
Certain scams convince victims to pay in gift cards or money transfers. Others trick victims with seemingly lucrative cryptocurrency investments, using modern spins on classic romance or fraudulent investment lures, and then transfer or sell off the invested funds before the victim can cash out (known as a “rug-pull”).
The underground economy must necessarily interact with the real-world economy to convert victim payments into legitimate currencies.
Laundering creates exposure and risk for cybercriminals because following the money is the most effective method for law enforcement to track down perpetrators and their suppliers.
While some cybercriminals operate out of nation-states that protect them from prosecution, others avoid detection by masking their execution. This is especially true for payouts: if illegitimate payments are frozen at any step, cybercriminals lose all attempted gains.
A suite of underground financial services exists to facilitate criminal payment processing. These services are designed to evade and hinder law enforcement’s ability to observe fraudulent transactions that process a victim payment into a successful conversion. They also obscure the relationship between a criminal organization’s key players and its supply chain.
Many criminal marketplaces provide underground banking services (e.g., the Dubai Bank), privacy coins, and even mule-as-a-service operations that facilitate laundering of various types of financial transactions.
Cybercriminal Laundering Techniques
Cryptocurrency Mixers
Also called tumblers, these services collect crypto transactions from multiple origin addresses, distribute them across multiple addresses, and then return equivalent cryptocurrency to the origin addresses to make cryptocurrency transactions hard to trace.
Crypto Chain Hopping
A method of making cryptocurrency hard to trace by rapid converting one cryptocurrency to another, e.g. Bitcoin to Monero.
Money Mules
Individuals recruited through fake work-from-home job ads to receive (and sell) stolen funds and transfer the money for a small take. Some mules are forced labor.
Game-based Money Laundering (“Loot Kits”)
Criminals purchase in-game currencies and virtual items (“loot boxes”) with stolen funds and resell what they purchase to other players for clean currency.
Smurfing
Criminals divide large sums into smaller (often microscopic) sums before depositing them across many bank accounts to bypass anti-money laundering (AML) reporting limits.
Following cryptocurrency is challenging. Law enforcement agencies must contend with countries that do not diligently pursue cybercriminals, due to a lack of resources, because they do not care about predominantly Western victims, or because they consider cybercrime part of their hybrid conflict strategies. These modern-day activities could be likened to seventeenth-century privateering.
Proactive Measures, Defenses, and Customer Cybersecurity Awareness
Defending against phishing and fraud in the CaaS era is a challenge for every organization. As with every business threat, risk assessments should prioritize deployments of proactive measures, defenses, and user cybersecurity awareness initiatives.
We have highlighted AiTM and MFA bypass as formidable threats. Organizations are adopting phishing-resistant MFA and applying session lifecycle controls. Where practical, they should enforce device allow listing, so users can authenticate only from approved devices.
Limiting what users can access once authenticated—applying the principle of least privilege—is another effective means to limit damage should an MFA bypass attack succeed. Many organizations allow users unbounded or minimally bounded access to servers, services and data. Organizations that complement authentication with authorization policies can restrict access based on the user’s role, which minimizes data at risk should a given user account be compromised.
The myriad ways that criminals misuse AI redefine the battlefield for every organization that is a target for fraud. Integrating AI into fraud detection programs is essentially a given. Whether implemented by a banking institution or outsourced to a trusted third party, AI may improve fraud prevention in several ways:
- AI can monitor unusually large or frequent cryptocurrency transactions or flag suspicious crypto addresses.
- LLMs can be trained on behavioral data to detect transfers characteristic of romance baiting or other investment scams.
- AI can be used to monitor the signals from transaction data, social media, and messaging channels where criminals hunt for prey and use these “cross-channel” signals to notify customers of suspected fraud or to protect customers by pausing transactions that would otherwise result in significant financial harm.
- Case studies by DigitalDefynd demonstrate that AI can be effective in anti-money-laundering efforts. A major benefit of AI tooling is its ability to trace criminal transaction flows through oceans of mostly legitimate financial exchanges. Money laundering succeeds when intentionally obfuscated transaction flows exhaust or confound investigators. However, an AI assistant can overcome the limitations of human investigators.
- Field success of deep learning or neural network (transformer) models illustrates that an advanced transformer model can improve credit card fraud detection.
- A generative AI-enabled fraud detection framework has proven successful in identifying suspicious activity in ATM, mobile app, web banking, and call centers.
Increasing cybersecurity awareness among executives, staff, and customers is challenging but essential.
BEC and whale phishing target executives and senior staff, especially those involved with financial transactions.
Security teams should prioritize ensuring these personnel understand these phishing threats and stay informed as they evolve. Transaction monitoring of key personnel, while potentially sensitive, may be appropriate for organizations that are determined to avoid potentially catastrophic loss.
Multi-party controls—for example, policies that require approval from an additional party or another department for transactions involving large sums—are a reliable means of preventing fraudulent transactions.
Most financial services staff understand technology well enough to benefit from modest-to-advanced technical training. Consider adopting compulsory online or team training sessions that focus on the frauds, scams, and social engineering techniques most used in BEC or corporate account theft accounts where the targets are executives or staff who process financial transactions.
Simulated vishing or phishing attacks conducted in financial services have had beneficial outcomes.
PhishKill reports that the average click rates for financial services organizations in the early stages of phishing simulation commonly fall in the 15-25% range, and click rates for finance or accounting staff departments fall in the 12-18% range.
Perhaps more importantly, StationX reports that conducting monthly simulated phishing attacks increased detection rates and shortened reporting times. A simulation program that prompts employees to report phishing rather than click a link can proactively prevent fraud losses.
Investment and romance scammers feast on older adults or socially isolated individuals, especially those they can identify as wealthy and liquid. These targeted populations are among the least tech-savvy customers of a financial service.
Explaining frauds and scams and how to avoid becoming a victim in simple terms may be most effective. Direct older adults to information sources that they trust—AARP, BBB, Chambers of Commerce—and government agencies: in the U.S., the FBI’s Internet Crime Complaint Center, Federal Trade Commission, National Institute on Aging, and Consumer Financial Protection Bureau. These sources make public service announcements and provide information about fraud hotlines, loss recovery, and victim support workshops that financial institutions can share with their customers or incorporate into their own awareness programs.
Take the first step in locking down your bank’s digital presence.
Register your official .Bank domain at Register.bank to protect your customers, eliminate domain spoofing, and meet the highest security standards for banking websites.
Or schedule a meeting with us to see how .Bank can help your bank.