Digital businesses facing the rise of AI-driven fraud are under immense pressure to enhance their protective measures. The emergence of synthetic identities and AI-generated documents poses a significant challenge to traditional verification processes, while automated attacks enable organized crime to exploit vulnerabilities rapidly.
Know Your Customer (KYC) checks have become particularly vulnerable, with skilled fraudsters increasingly using AI to impersonate genuine customers. “Currently, KYC is the most targeted layer, where real-time document generation and deepfake face-swapping are becoming remarkably sophisticated,” stated Volodymyr Todurov, CEO of Frogo, a provider of fraud prevention and risk management solutions.
However, implementing stringent controls can lead to unintended consequences. While aggressive measures might block suspicious transactions, they can also result in the denial of legitimate payments and delayed withdrawals, pushing customers to seek services elsewhere. As a result, businesses may lower their reported fraud rates but risk losing valuable clientele in the process.
Frogo, which integrates device intelligence, automated risk scoring, and real-time alerts, suggests that companies rethink their approach to fraud prevention. “Operating with a ‘guilty until proven innocent’ mindset and relying on aggressive, automatic blocks,” Todurov cautions, “treating legitimate players like suspects completely destroys the customer experience.” This approach raises a broader question about anti-fraud effectiveness: Does it contribute to profitability when factoring in the costs of prevented losses, operational overhead, and the impact on genuine customers?
To evaluate the business implications of fraud prevention strategies, Todurov likens it to a “revenue optimization engine,” measured through three interconnected dimensions: direct loss prevention, operational efficiency, and customer retention.
Direct loss prevention involves tracking the financial losses attributed to fraud, including fraudulent withdrawals, and assessing whether these figures are declining. However, the resources required to achieve this reduction are equally important. If managing routine checks requires constant manual oversight, businesses may find it difficult to handle increasing customer volumes without continually expanding their risk management teams.
“Without a nuanced evaluation model, you risk blocking your highest-revenue players,” warns Todurov.
Another metric to consider is the ratio of monthly active users per fraud analyst, along with the time needed to complete investigations. By automating routine decisions, specialists can focus on more intricate threats.
The third component examines the experience of legitimate customers. Todurov proposes implementing “automated withdrawals built on the presumption of innocence,” suggesting that smoother transaction processes can lead to enhanced trust and promote long-term customer value. Delays and unnecessary checks impose costs that may not be captured in fraud assessments.
These dimensions must be analyzed collectively. An effective system may accurately flag certain fraudulent activities while overlooking significant fraud elsewhere, or may diminish losses through overly rigorous controls that drive away real customers. The financial result hinges on understanding what the system fails to identify and the costs associated with its interventions.
High-value customers exemplify how stringent controls can misfire. Their behavioral patterns often differ significantly from those of average users; frequent travel can cause sudden location changes, large deposits can create unexpected spikes in activity, and the use of multiple payment cards adds further nuance. These signals may resemble fraudulent behavior, such as attempts to test stolen cards, leading businesses to potentially reject some of their most valuable customers.
“Without a nuanced evaluation model, you risk blocking your highest-revenue players,” Todurov reiterates.
Frogo argues that atypical behavior should not automatically trigger a block but should prompt informed assessments instead. Depending on the situation, unusual signals may warrant intervention from support or retention teams rather than an outright denial.
Recognizing these distinctions relies on specialists who understand the business and its customers, enabling them to determine which signals require action. This expertise also influences where AI can enhance decision-making versus areas where established rules remain more effective.
“AI isn’t a silver bullet,” says Todurov. “Its success depends on specialists who know precisely where machine learning adds value and where tried-and-true, deterministic logic is still superior.”
Gathering essential information from a user’s first interaction is crucial for these specialists. Frogo emphasizes the importance of starting with the device used to access the service and refining risk assessments as the user’s activity evolves. “It all begins with device fingerprinting, where we analyze hardware and software parameters to verify user legitimacy and detect bots, spoofed devices, or masked origins,” Todurov explains.
Frogo’s standalone Device Fingerprinting tool allows businesses to enhance their current anti-fraud measures independently of their broader platform. Todurov highlights companies with basic in-house rules but limited device recognition as key beneficiaries. This tool adds context to signals like sudden location changes, aiding analysts in determining whether the activity aligns with behavior expected from a returning customer or if it should prompt further scrutiny.
“Anti-fraud tools should be viewed as core data engines,” he asserts.
The tool evaluates over 300 client-side signals, producing identifiers that serve two important purposes: maintaining continuity within a single domain and recognizing the device across various brands within a portfolio. Additionally, it supplies context, including IP risk, VPN and proxy information, browser specifics, screen resolution, and hardware characteristics. Other signals encompass time zone mismatches, remote desktop tools, and virtual machine detection.
Within Frogo’s comprehensive platform, any actions such as registration or transaction attempts are checked against the business’s risk parameters. Todurov mentions that this system responds in less than a second, explaining which risk triggers were activated.
The benefits of this information extend beyond mere threat identification; it can also help businesses recognize beneficial customer behaviors. “Anti-fraud tools should be viewed as core data engines,” asserts Todurov. “Instead of only scoring bad behavior, you can define triggers that identify high-value, ‘ideal’ player behavior.”
Through its real-time alerting system, Frogo directs negative risk signals to analysts while sending positive behavioral notifications to marketing and VIP teams via webhooks or Telegram, enabling various departments to react appropriately to the same underlying activities.
Applying this method in new markets requires ongoing adaptation. Customer behaviors, payment preferences, and attack strategies differ, necessitating revisions to rules that may have been effective in one environment but not in another.
The overarching commercial goal is to effectively manage genuine threats, minimize unnecessary work, and maintain seamless experiences for legitimate customers as businesses grow.
“The winners will combine a highly skilled fraud team with agile technology, viewing anti-fraud through a growth mindset,” Todurov summarizes.
