A girlfriend who doesn't exist and $160,000 lost: AI turns loneliness into the new fraud vector

By La Nación · October 12, 2026.

Mario (not his real name), 84, a widower with adult daughters, spent months in a WhatsApp relationship with "Luciana," a young, successful woman with social media accounts and even a LinkedIn profile. According to La Nación's account, based on a talk by Gabriel Zurdo, CEO of the consultancy BTR Consulting, Luciana did not exist: she was an AI capable of chatting with thousands of victims at once. Over four months, she persuaded him to invest $160,000 in a platform that promised strong returns.

The mechanics are familiar. The app and website where Mario watched his money grow were fake. When he tried to withdraw funds, he was asked for additional deposits to "release the payment," one after another. When he refused, his account was closed and Luciana blocked him. Part of the money was handed over in cash at an unlicensed exchange house (a "cueva") in Buenos Aires' financial district, the City, where it was converted into cryptocurrency. According to the newspaper, the investigation is still ongoing, and that location is the thread investigators hope to pull to try to recover it.

The most revealing detail is not the loss. Zurdo says the hardest part was that Mario did not believe his daughters when they told him Luciana did not exist. The daughters found out by chance, when he began taking on debt in another venture so he could keep sending money.

The technique has a name: *pig butchering*. It consists of building trust or romance over weeks, inducing the victim to deposit capital into fake platforms that simulate successful withdrawals at first, and then freezing the funds and extorting the victim to release them. BTR Consulting counts it among the three key types of fraud targeting individuals.

The numbers, according to the data cited in the article: the 2025 survey by Argentina's National Criminal Information System (SNIC) records 50,155 online-enabled scams and frauds in the country. That is almost one online case for every eight in-person robberies. Added to traditional scams (72,953), fraud now accounts for almost a third of property crimes. BTR adds that more than 50% of incidents reported at Argentine institutions are phishing or cyberfraud. These figures come from the consultancy and from official sources; the article itself acknowledges that underreporting leaves a wide gap between statistics and reality.

**Our take.** What generative AI changes is not the romance scam, which is old, but its economics. Previously, maintaining hundreds of fake relationships required call centers and human operators. Now a language model can sustain intimate conversations, at any hour and with memory of family details, at almost zero marginal cost. Fraud stops scaling with people and starts scaling with compute. It is the same pattern of automated deception we see across the rest of cybercrime: the real short-term danger of AI is not some distant superintelligence, but this industrialization of fraud.

There is a caveat worth noting. The article states that Luciana is an AI, but does not explain how this was verified. These schemes often involve human operators supported by automated tools. For the reader, the distinction matters less than the conclusion: the other party may be synthetic and the bond can still feel real.

It is also an uncomfortable story about loneliness. The attack vector is not a technical vulnerability but a human need for companionship. Isolated older people are the perfect target, and there BTR's advice on defensive AI, zero trust and supply chain security, designed for businesses, is of little help. For individuals, the useful defenses are simple: be wary of any relationship that leads to investments, do not move money to platforms that cannot be independently verified, and talk to family before sending funds. In general, verification by a second trusted person remains the most effective barrier.

The case also points to unfinished business for platforms and regulators. If a system can converse with thousands of people at once, detecting coordinated patterns is the job of whoever hosts those conversations. Security Minister Monteoliva acknowledged, according to the article, a "major challenge" in tools, training and communication. It is a reasonable response, though the article does not yet mention concrete measures.

The bottom line is the one we keep returning to. In the short term, the transition brings real and unequal harms, and older people living alone pay first. In the long term, the same technology that today simulates affection in order to steal can offer genuine companionship, support and care, and defensive AI can neutralize a large share of these frauds. Getting there requires treating automated fraud as a public safety problem, not as an anecdote about naive victims.