
Digital Content

- Unlimited access to peer-contribution articles and insights
- Global research and market intelligence reports
- Discover iNFRont Magazine, an NFR publication
- Panel discussion and presentation recordings



- Online fraud costs businesses an average of $10.6 million annually
- Ravelin finds 76% of firms feel pressured to
approve suspicious refunds
- Feedzai’s OpenL2D framework shows AI-human
collaboration outperforms either alone
- Generative AI is fuelling deepfakes and
synthetic identity attacks
- Experts warn against isolating fraud teams from
product and customer functions
- Data silos limit fraud teams’ ability to detect
and act in real time
- Unified data strategies reduce false positives
and enhance user experience
- AI can scale detection, but humans are essential
for context and ethics
- Fraud must be integrated into product,
marketing, and operational workflows
- Firms that treat fraud as strategic
infrastructure, not overhead, will gain trust and reduce losses
Online fraud is no longer a background compliance issue – it’s a strategic fault line exposing deep dysfunction inside financial institutions.
According to Ravelin’s Global Fraud Trends 2025 report, online merchants now lose an average of $10.6 million annually to fraud. Yet what the research reveals more starkly is not just the volume of loss, but the widespread inaction driving it.
“The biggest misconception is that fraud prevention degrades the customer experience,” said Martin Sweeney, CEO of Ravelin. “But the right intelligence doesn’t slow down trusted users – it clears the way for them.”
Ravelin’s study of over 1,400 professionals found that 76% of businesses feel pressured to approve refunds even when fraud is suspected.
This dynamic, most visible in retail and travel, is now infecting financial services, where seamless experiences are prized and friction is feared.
Sweeney warns this is a dangerous mindset. “Too many businesses dismiss fraud as a cost of doing business,” he said. “That’s a false dichotomy. It’s entirely possible to deliver secure digital journeys that also protect your bottom line.”
The problem is compounded by another reality: fraud teams often operate in isolation. They lack real-time access to data from customer service, product, and operations – making it nearly impossible to interpret user behaviour holistically.
“Your data already knows who to trust,” Sweeney said. “But without infrastructure to act on that insight, institutions are flying blind.”
Feedzai’s latest research takes this further, unveiling OpenL2D, a machine learning framework developed to help AI systems defer to human judgment in high-stakes decisions.
Based on 30,000 real-world fraud cases and published in Nature Scientific Data, the system allows banks and fintechs to evaluate the balance between automation and human oversight.
“Fraudsters are using generative AI to craft synthetic identities and near-undetectable scams,” said Jas Anand, Senior Fraud Executive at Feedzai.
“AI must be the first line of defence, but people provide context, especially in ambiguous or ethically sensitive cases.”
Anand argues that hybrid intelligence – not automation alone – is the future of fraud defence. “It’s not about replacing teams,” he said. “It’s about giving them better tools and the time to think strategically.”
However, the greatest threat may not be fraudsters at all. It’s internal fragmentation. When browsing history, payment records, customer support logs, and device data are analysed separately, even the best fraud models miss the mark. High-value customers get flagged. Real criminals slip through.
“Too many institutions still assess transactions in isolation,” Anand said. “They need to shift from reactive alerts to contextual, real-time decision-making.”
The urgency is clear. As digital services expand and AI advances, fraud is evolving from a technical problem to a strategic battleground. The firms that win will be those that reframe fraud prevention as a pillar of customer trust and growth—not a barrier.
That means repositioning fraud as a leadership priority. It means tearing down data silos. It means embedding fraud controls into product design, marketing, and service flows.
And it means engaging with cross-platform intelligence networks to detect patterns that no single firm can see alone.
“Fraud isn’t a back-office issue anymore,” said Anand. “It’s a front-line test of a company’s maturity – and its values.”