Romance scams aren’t new, but their scale, sophistication and emotional precision have transformed dramatically in recent years. Today’s scammers build relationships that feel real, nurturing trust over weeks or months before exploiting it — leaving victims financially devastated and emotionally shattered. These scams don’t begin with a suspicious email or an unexpected phone call. They begin with something far more disarming: connection.
A Modern Love Story: Manufactured Intimacy
The profile picture presented a handsome man in his 50s with salt-and-pepper hair, a warm smile, dressed in business casuals. “David” reached out to Linda on a social media platform, commenting on a photo she had posted from her book club. David was charming, articulate and claimed to be a civil engineer working on a large construction project in the UAE. Within days, they had moved their conversations to an encrypted messaging application. Linda’s day started with “Good morning, beautiful” texts from David and ended with late-night messages.
Over three months, David became Linda’s confidant. He learned more about her life, her recent divorce, her worries about retirement savings and her dream of travelling and seeing the world. David, like a true friend, listened initially, offering emotional support. Nothing about the relationship felt rushed. After a few weeks, he suggested an investment opportunity that had helped him earn significant returns. He suggested starting with a small amount, between $1,000 and $2,000, to test the waters.
Linda wired the money from her bank account to a cryptocurrency trading platform. Her investments started to show impressive returns just as David had suggested. Emboldened by her initial success, she invested more and more. In the next six months, Linda transferred a total of $127,000. In this process, she had liquidated her 401(K) and started a home equity line of credit. When she tried to withdraw her profits, the platform demanded a 30% tax payment. That’s when Linda realized that the platform was fake.
What she discovered later made it worse. Linda’s daughter brought it to her attention that David's photos had been stolen from the profile of a marketing executive in Texas who had never even heard of her. A simple image search could have revealed the deception, but she had no reason to be suspicious. When Linda’s daughter reported the platform, it took three weeks to get a response, and by then the fraudster’s account had been deleted and recreated under a new name, ready for the next victim.
This pattern is familiar to many victims of romance scams. Linda’s story is not a single documented case. It is a composite synthesized from multiple accounts of victims who have come forward and reported such scams to law enforcement and consumer protection agencies. Linda’s story mirrors thousands of others.
In 2024, the FBI’s Internet Crime Complaint Center received reports from 147,127 people over the age of 60 who were victims of romance scams, resulting in direct losses of hundreds of millions of dollars – including $1.8 billion in cryptocurrency investment fraud affecting older adults.
At multiple points in Linda’s journey, intervention was possible. The social media platform, the messaging app or the crypto trading platforms could have acted and stopped the scam. But none of these interventions happened. The only institution facing scrutiny was Linda's bank — simply for processing six authorized wire transfers, made with proper credentials, to a legitimate cryptocurrency exchange.
Accountable and Under-Equipped
In many ways, banks are at a disadvantage when it comes to addressing and preventing romance scams. This isn't just a practical or operational challenge - it is being codified into regulation. As governments rush to address the rise in scams, regulatory frameworks are emerging that, despite best intentions, disproportionately burden financial institutions while leaving the platforms where scams originate largely untouched.
Even Australia’s Scam Prevention Framework, widely viewed as the most progressive global attempt to tackle consumer scams, exposes a structural limitation. Under the current proposal, entities such as dating apps, online marketplaces, cryptocurrency exchanges, non-bank payment providers and email services remain outside the scope of mandatory controls, despite being common entry or escalation points in romance scams.
“What the SPF demonstrates is that even the most advanced regulatory frameworks still struggle to address scams end-to-end. When key platforms and payment providers are excluded, and compliance is self-certified, the practical burden of intervention and reimbursement will inevitably fall on banks more often, at a point long after the scam relationship has already taken hold,” says Ken Palla, former Director at MUFG Union Bank.
Even if we accept this regulatory burden, which is debatable given where scams originate, banks face a more fundamental problem: the fraud detection systems that are being used currently were never designed to catch romance scams. Most institutions are fighting a new threat with dated tools optimized to only detect account takeovers and other forms of unauthorized fraud.
Transaction fraud detection systems are trained to identify deviations from established behavioral baselines. The underlying assumption is that fraud looks different from normal behavior. This works brilliantly for account takeovers (ATOs) or other scams that rely on smash-and-grab approaches and are executed within a few minutes. Sudden changes in behavior or anomalies are quickly flagged, and the fraud is easily identified and prevented. Romance scams work quite differently. As we see in Linda’s experience, rather than create sudden anomalies, scammers gradually shift victims' normal behavior over months. By that time, both the victim and the fraud detection systems have fallen under the scammer's spell.
From Anomalies to Continuity: Rethinking Scam Detection
Beyond the temporal mismatch, romance scams present two additional detection challenges. Firstly, these are authorized transactions. The victim is using their own device from a known location with correct credentials. Unlike ATO, there are no device or IP anomalies here to flag suspicious behavior. Additionally, and most critically, there is no shared intelligence across financial institutions or cross-industry groups. A scammer who scammed Linda at Bank A can now move to another bank to look for a new victim with no warning flags.
Detecting romance scams requires a fundamental shift in our approach from anomaly detection to continuity detection. Instead of looking for anomalies in transaction behavior, we need to look for breaks in behavioral continuity.
"Financial behavior is inherently sequential, contextual, and path-dependent," explains Danny Butvinik, Chief Data Scientist at NICE Actimize. "Risk is not defined by extremity, but by incompatibility — a behavioral continuation that is statistically and semantically inconsistent with the latent dynamics inferred from historical context. Scam scenarios often involve gradual manipulation, rehearsal phases and delayed execution, where individual actions appear legitimate until viewed in sequence."
Machine learning models built on these principles, particularly deep learning architectures designed for sequential pattern recognition, can track behavioral evolution over extended timeframes and identify the subtle shifts in transactional behaviors that traditional systems miss.
But technology alone isn't enough. When a scammer successfully deploys tactics at Bank A, Bank B remains completely unaware. The Federal Reserve's Scam Information Sharing Industry Work Group has recognized this gap, recommending enhanced real-time information sharing and standardized data formats across financial institutions. New vendor-led consortia are emerging to operationalize this vision — creating networks where banks can share scam patterns, suspicious beneficiary lists and cryptocurrency wallet addresses associated with romance fraud. These frameworks allow pattern recognition across the industry, not just within individual institutions, making it harder for scammers to simply replicate successful tactics with new victims at different banks.
Where We Go From Here
Yes, banks have been dealt an unfair burden. But if we’re going to be held accountable for preventing romance scams, we can’t keep using tools from 2006 to fight the threats in 2026. The technology exists, andfraud teams are eager to do more. What's needed now is execution - before another 147,000 victims are added to the statistics.
Modernize Your Fraud Prevention Program