AI Romance Factories Targeting Americans

Scammers are mass-producing fake romance with American tech—and they only need days to make you trust them.

Story Snapshot

U.S.-made artificial intelligence tools help scammers scale love fraud across languages.
Traffic from Myanmar scam compounds rides on U.S. internet providers.
The Treasury sanctioned Southeast Asian networks that target Americans.
Americans lost at least $10 billion to these scams in 2024, and losses are rising.

How the love con went from catfishing to a factory line

Associated Press and FRONTLINE found that scammers use American-made artificial intelligence models, including ChatGPT and Gemini, to build software that writes messages, translates in real time, and tracks results across dozens of languages. This turns a smooth talker into a 24/7 call center. Workers follow scripts, refine lines, and rotate shifts. The goal is speed. Hook a lonely target in days, then move them toward crypto or wire transfers. The script is cold, but the tone feels warm because the machine polishes it.

One weakness in the record deserves attention. The investigation describes the software’s abilities and how scammers bought and used it, but it does not publish raw logs that link named scam accounts to specific artificial intelligence model accounts. That gap does not blunt the larger pattern. It shows why policy must push for audit trails. When tools can charm at scale, we need receipts, not just inferences, to back penalties that stick in court.

The pipes that carry the heartbreak

American internet firms play a bigger role than most expect. The same probe traced more than 200,000 device connections from four Myanmar compounds linked to sanctioned entities. One in five signals from those devices rode on U.S.-registered internet service providers, including Cogent Communications and AT&T, and on cloud services from companies such as DigitalOcean and Oracle. This is not a niche corner of the web. It is the main road system. Criminals exploit the best highways because they work.

Washington’s response shows the danger is real and known. The Department of the Treasury’s Office of Foreign Assets Control sanctioned dozens of people and businesses tied to romance and cyber fraud networks that target Americans. Sanctions make it harder to move money and buy services. But sanctions alone do not rewire the internet. As long as routing stays cheap, fast, and global, bad actors will keep hopping carriers and clouds to dodge bans. That is the structural fight we must win.

Why your neighbor keeps falling for it

The modern “pig butchering” playbook is now tuned for Western victims. The United States–China Economic and Security Review Commission reports a mutation that targets foreigners with tailored chats, slick images, and staged video calls. Translation software and messaging apps remove language barriers and speed the bond, then the pitch lands: crypto or investment “opportunities” that drain savings. This is not about gullibility. It is about a machine-tested script that exploits normal hope and trust under time pressure.

The losses tell the story. Americans lost at least $10 billion to Southeast Asia–based scams in 2024, and projections say 2025 will be worse. Families lose retirements. Small-town charities lose donors. Local police struggle because offenders sit behind barbed wire in border zones. Federal agents face a maze of shell firms and cross-border data requests. The criminals run shifts; the victims run out of time. That imbalance favors crime until policy closes the gap.

Fix the incentives, shrink the scam

Conservative common sense starts with accountability. If a U.S. company’s network carries a known scam compound’s traffic, the company should prove it ran risk checks and blocked what sanctions require. The Associated Press analysis shows an outsized role for U.S. carriers in Myanmar compound traffic. Stronger know-your-customer rules for cloud and communications, time-bound takedown orders, and published transparency reports would raise the cost of looking away. Honest firms will adapt. The worst actors will lose easy access.

Policy should also demand traceability from artificial intelligence service providers when lawfully required. Open audit logs and verified developer identities would let investigators match scam apps to model usage, with due process. The Associated Press report flagged the lack of direct model account logs in public. That is solvable. Build privacy-safe tokens for legitimate users, but keep legal keys for subpoenas tied to sanctions and fraud cases. Freedom thrives when law and order can act fast.

What to do before the next message arrives

Guard your pace. Real trust takes time. Scammers rush because delay breaks the spell. Refuse to move chats off mainstream platforms, decline crypto or wire “investments,” and video call with a simple shared gesture that defeats deepfakes. Share suspicions with a friend before you send a dollar. If you feel sudden urgency mixed with flattery, hit pause. The other side has scripts and software. Your edge is time, second opinions, and the courage to say no.