At Mar-a-Lago’s First Policy Institute gala, where celebrities like Sylvester Stallone, Elon Musk, Pam Bondi, Argentina’s Javier Milei, and Donald Trump mingled, a different kind of story was making waves.
A persistent local reporter couldn’t stop talking about what they called “The Fabiola Fraud.” The case that inspired FABIOLA—a predictive AI model designed to uncover marriage fraud networks—was described as “the most brazen attempt at marriage fraud in Florida history—an ideal case for the new administration to make an example of.”
But this isn’t just one scandal—it’s a glimpse into something much bigger.
A Marriage Fraud with a $20,000 Price Tag
At the center of this controversy is Fabiola Carisimo, a Miami realtor accused of paying $20,000 to orchestrate a fraudulent two-week marriage to secure U.S. residency. On paper, the marriage was presented as legitimate, with affidavits and documentation to back her claim. But investigators quickly uncovered inconsistencies, raising questions about the authenticity of the arrangement.
The “Husband” Who Wanted More
In a twist that further complicates the case, Julimir Marrero Perez, Carisimo’s “husband,” allegedly attempted to renegotiate the terms of their deal after the marriage. According to sources familiar with the situation, he demanded additional money—on top of the original $20,000—to maintain the facade. This move not only highlights the transactional nature of the relationship but also raises suspicions about how frequently similar renegotiations occur in Miami’s marriage fraud ecosystem.
The “Wife” Who Never Let Him Move In
Adding to the suspicion is the fact that Carisimo’s husband never even moved in with her. Despite claims that the marriage was legitimate, there was no evidence of cohabitation—a critical factor in proving a genuine relationship. Instead, their arrangement appeared to exist solely on paper, reinforced by the questionable affidavits and documents submitted to immigration authorities.
The Judge Who Had Enough
The case came before Judge Abby Cynamon, who dismissed it not once, but twice, demonstrating her commitment to ensuring that cases brought to court meet the necessary evidentiary standards.
Judge Cynamon’s actions reflect the crucial role of judicial discretion in navigating complex and often high-profile cases, particularly in a city like Miami, where marriage fraud allegations frequently surface.
By thoroughly examining the facts and dismissing a case lacking sufficient evidence, Judge Abby Cynamon reinforced the importance of maintaining integrity in the legal process while protecting the courts from being misused in potentially fraudulent situations.
An Eight-Year Boyfriend
Further complicating the narrative is Carisimo’s eight-year relationship with a long-term boyfriend from Argentina who frequently visited her and, which she maintained throughout the short-lived marriage. This revelation casts significant doubt on the marriage’s validity and underscores the audacity of attempting to balance an ongoing relationship while staging a fraudulent marriage for immigration purposes.
Abuse Claims
Fabiola Carisimo’s marriage on July 24, 2023, lasted just two weeks, with the couple separating on August 8, 2023. According to Carisimo: On August 10, she received a text from Marrero asking her not to discuss their issues with others, and the following day, she filed a restraining order citing incidents dating back to May. Despite the brief marriage and no evidence of cohabitation, she waited exactly eight months to the day of their wedding—March 24, 2024—to file for divorce.
Given the transactional nature of the $20,000 marriage, the absence of police reports for the alleged incidents, and the precise timing of her legal actions, critics argue that these filings may have been strategically planned rather than reflective of genuine abuse or marital intent. The case highlights broader concerns about the potential misuse of legal protections, particularly in situations involving suspected marriage fraud.
A Pattern of Contradictions
From the payment for her marriage to the eight-month delay in filing for divorce, Fabiola Carisimo’s actions reflect a pattern of calculated moves that raise significant doubts about the authenticity of her abuse claims. While abuse allegations must be taken seriously, this case serves as a reminder of the importance of thoroughly evaluating evidence to protect the integrity of the justice system for genuine survivors.
How AI and Social Media Reveal Hidden Networks of Marriage Fraud
VAWAWatch is actively developing a predictive AI model designed for collaboration with key stakeholders. The predictive AI model called FABIOLA (Fraud Analytics and Behavioral Insights for Optimizing Law Applications) is poised to support agencies like DHS in uncovering fraud networks, leveraging innovative approaches to combat marriage fraud.
Why FABIOLA?
Our predictive AI model, FABIOLA (Fraud Analytics and Behavioral Insights for Optimizing Law Applications), is named after the Fabiola Carisimo case, a striking example of the complexities and challenges involved in uncovering marriage fraud. This case underscores the need for tools that can sift through patterns, social networks, and inconsistencies to identify fraud while protecting the integrity of the legal system. Naming the system FABIOLA serves as a reminder of why innovative solutions are critical in tackling systemic issues like marriage fraud—one case at a time. Designed to uncover marriage fraud at scale, the system analyzes public data to identify high-probability offenders and their enablers.
In the case of Fabiola Carisimo, the FABIOLA algorithm scrutinized her 735 followed accounts on Instagram, using advanced regression models to evaluate patterns. The findings were startling:
- 12.9% (94 accounts): Likely participants in marriage fraud schemes based on specific behavioral markers.
- 9% (66 accounts): Accounts showing signs of significant awareness of these schemes but no direct involvement.
Followers Are Mostly Fake—But Not Irrelevant
While Carisimo’s followers list wasn’t central to our analysis, it’s worth noting that 80% appear to be fake or inactive accounts. Metrics such as low engagement rates, bot-like usernames, and geographically mismatched profiles suggest she inflated her follower count to project credibility.
However, the 20% of real accounts remain under scrutiny. Even among fake followers, the algorithm detects patterns of interaction and possible traces of real-world fraud networks. No one gets a free pass just because they’re hiding behind a wall of bots.
The Algorithm’s Approach: Beyond the Obvious
Our model doesn’t just rely on superficial indicators like shared photos or geotags. It digs deeper into less-discussed factors, some of which are derived from international studies, such as those pioneered in the UK to combat similar fraud. Here’s what we analyze:
- Demographics:
- Age disparities: Accounts linked to fraudulent activity often show unusual age gaps between spouses (e.g., a 45-year-old sponsor with a 21-year-old spouse).
- Gender patterns: Certain behaviors tend to cluster along gender lines, such as females posting luxury items after quick marriages or males sharing vacation-style photos without their “spouse” present.
- Timing:
- Marriage filing timelines: Patterns where a marriage follows visa denials or near-expiration of residency statuses.
- Account activity spikes: Posting frequency surges around weddings but little to no interaction afterward.
- Cultural Mismatches:
- While not determinative, flagged accounts often show visible cultural differences between couples, paired with behaviors inconsistent with genuine relationships (e.g., no shared friends or family in photos).
- Financial Indicators:
- Sudden changes in lifestyle: Posts showing luxury vacations, designer items, or unexplained wealth shifts often raise red flags.
- Overlapping recruiters, immigration consultants and/or paralegals: Accounts that appear connected to multiple high-risk relationships are tracked closely.
- Second-Degree Connections:
- Our network mapping feature detects clusters of accounts that frequently interact with known fraud participants. This is how Carisimo’s connections were flagged as high-risk, even when they attempted to appear unaffiliated.
Bias in Fraud Detection: A Necessary Reality
Unlike other systems hampered by fears of bias accusations, VAWA Watch embraces bias as a strategic tool. Our algorithm is admittedly biased, and it doesn’t pretend that fraud is equally distributed across all demographics, cultures, or behaviors.
- Age and financial patterns matter: A 50-year-old sponsor with a sudden young spouse is far more likely to be fraudulent than two 30-year-olds marrying after years of documented dating.
- Cultural data is relevant: Certain fraud markers appear more frequently in specific communities or immigration pathways, and ignoring these trends undermines the system’s effectiveness.
- Behavioral profiling works: Whether it’s posting patterns, tagged locations, or shifts in lifestyle, the algorithm unapologetically uses these biases because they are based on hard data, not politics.
“We aren’t afraid to follow the data, even when it’s uncomfortable,” one VAWA Watch researcher said. “Our goal isn’t to be politically correct or to play social engineers—it’s to give DHS the tools to dismantle these underground marriage networks in a cost-effective way.”
For those connected to Carisimo’s network, this analysis serves as a stark warning. AI technology doesn’t stop at the surface—it digs into the metadata, behavioral patterns, and even hidden connections. Deleting photos or locking down accounts won’t erase the digital trail. Once flagged, even private accounts leave enough traces for secondary analysis.
For high-risk accounts flagged in Carisimo’s network, scrutiny by both AI and federal investigators is inevitable. If you’ve interacted with her on social media and know about—or participated in—fraudulent activity, consider this your warning: the walls are closing in.
What This Means for the Future
This case isn’t just about one person—it’s about redefining how fraud is detected and prosecuted. VAWA Watch’s predictive AI model is setting a new standard, and while some may critique the bias it employs, the numbers speak for themselves. Fraud is being caught faster, networks are being dismantled, and those connected to schemes like Carisimo’s are learning the hard way: social media isn’t as private as they think.