Fraud continues to pose a critical risk to financial institutions worldwide. Despite advances in technology and regulation, many organizations combat fraud in isolation, leaving systemic gaps that criminals exploit. To effectively address this challenge, institutions must move from siloed defenses to collaborative, ecosystem-wide approaches.
On 09 September 2025, K2 Integrity and Coenibium Advisors brought together a panel of experts that included John Kidd, senior director at K2 Integrity; Ajit Tharaken, CEO at Consilient; Arion Petasis, founder and CEO at Coenobium Advisors; Rick Hoehne, fraud specialist at Coenobium Advisors; Constantin von Altrock, founder of Fraud Averse; and moderator Richard Hills, head of the UAE office and senior managing director at K2 Integrity. They discussed the challenges of siloed fraud management, the role of collaboration, the promise of artificial intelligence (AI) and federated machine learning, and the evolving regulatory environment. The conclusion is clear: unified, collaborative approaches are essential to safeguard the financial ecosystem. Click here to view a recording of the event.
The Challenge of Siloed Fraud Management
Fraud is not an isolated event but a systemic threat. When intelligence about fraudulent activity is not shared, criminals can simply redirect attacks from one institution to another. For example, a phishing campaign thwarted at a major bank can be quickly redeployed against smaller banks with fewer defenses, leading to avoidable losses and reputational damage. This fragmented approach results in inconsistent detection capabilities, delayed responses, and preventable losses across the financial system.
The reality is that the entire ecosystem pays the price when fraud escalates. The next crisis may not come from credit or liquidity—but from unchecked fraud and scams. To avoid this, institutions must recognize their interdependence and embrace solutions that allow for collectively informed responses to new and emerging fraud threats.
Measuring Success in Fraud Prevention
Defining effective metrics is difficult, since success often means preventing events before they happen. A balanced approach includes keeping an eye on the following elements:
- Fraud Losses—Tracking reductions in both absolute losses and relative exposure.
- Operational Costs—Managing the volume of alerts and the resources required for investigation.
- Customer Experience—Ensuring controls do not undermine trust or create excessive friction.
Monitoring the rate of change in losses and emerging risks provides additional insight into whether controls are keeping pace with evolving threats.
Building Collaborative Ecosystems
Building collaborative ecosystems requires both cultural and structural changes. At the organizational level, governance must prioritize fraud prevention, with boards actively engaged and informed. Institutions can share insights into risk events and fraud typologies without compromising sensitive internal data, creating collective resilience. Industry associations and joint initiatives, such as shared transaction monitoring programs, illustrate how collaboration can strengthen defenses.
Regulatory frameworks are also shifting toward shared responsibility. For example, new requirements in global markets now place liability for fraud events on both sending and receiving institutions. This regulatory direction underscores that collaboration is no longer optional—it has become a necessity. We also must not forget the importance of public-private partnerships. Although the best approach has yet to be determined, it is imperative to develop some form of collaboration between public regulatory bodies and international financial institutions that allows for more effective mitigation of fraud. Technological innovation may hold the answer.
Technology as an Enabler
Modern fraud prevention increasingly relies on advanced technologies that enable collaboration, real-time detection, and proactive intervention. Two particularly impactful innovations are federated machine learning and generative AI.
Federated Machine Learning (FML): This approach allows institutions to retain data locally while contributing to shared AI models. The aggregated insights strengthen fraud detection across the ecosystem without compromising privacy or violating cross-border data restrictions. It is a panacea for sharing intelligent insights without sharing information. Benefits include:
- Privacy-preserving collaboration—Institutions comply with data protection and cross-border restrictions while still contributing insights.
- Scalable learning—Smaller institutions with limited transaction volumes benefit from the diversity of data across the network.
- Improved accuracy—Broader data inputs help reduce false positives and detect previously unseen fraud typologies.
- Operational efficiency—Shared insights accelerate detection and shorten the time window during which criminals can exploit vulnerabilities.
Challenges include ensuring data quality, preventing bias or poisoning of models, and maintaining computational resources. However, with strong governance and technical safeguards, FML presents a scalable and secure path forward.
Generative AI: Traditional fraud detection systems flag anomalies based on past behavior or scoring models. Yet newer scam types—investment fraud, romance scams, impersonation schemes—often manipulate customers directly, making them unwitting participants. Generative AI expands the toolkit by:
- Engaging customers in real time—Automated dialogues can prompt customers during high-risk transactions (e.g., unusual crypto transfers) to validate intentions and surface red flags.
- Contextual education—Conversations can help customers recognize scam tactics themselves, reinforcing awareness while providing protection.
- Adaptive decision-making—AI can weigh both incriminating and exonerating evidence, giving institutions a more nuanced view of whether to allow, delay, or block a transaction.
These technologies are most effective when combined with modern fraud platforms. Legacy, siloed systems are ill-equipped to integrate advanced AI models or scale across multiple use cases. By adopting open platforms with robust data strategies, institutions can fully leverage FML and generative AI to expand detection capabilities, reduce false positives, and deliver stronger customer protection.
Balancing Innovation and Compliance
Regulatory expectations are evolving alongside technology. Financial institutions must align innovative fraud detection methods with compliance obligations, ensuring that privacy, governance, and oversight are preserved. Hybrid approaches that combine AI-driven prioritization with mandated investigations can both optimize resources and meet regulatory requirements.
To modernize effectively, institutions should:
- Take a long-term and consultative approach towards updating their outdated, siloed fraud systems.
- Adopt enterprise-wide, open fraud prevention platforms.
- Implement modern data strategies that integrate collaborative insights.
Conclusion
Fraud prevention cannot be managed in isolation. The financial ecosystem is only as strong as its weakest participant. By embracing collaborative governance, leveraging advanced technologies like federated learning and generative AI, and aligning with evolving regulations, institutions can move toward a more resilient fraud prevention framework.
Speed, intelligence, and unity are the industry’s greatest assets in the fight against fraud. The path forward lies in collective action that strengthens the system as a whole.