Safeer Mohammed Koorimannil had a task: to make victims fall in love within four days, and he had many victims. Trafficked to a scam center in Myanmar, he impersonated Ella, a 28-year-old Singaporean woman. During his shifts, Koorimannil engaged with more than 100 people using dozens of profiles while supervisors monitored with electric batons. In a month, records he sent to The Associated Press show he targeted 50,000 victims from 17 countries, including a tailor in Kurdistan, a pastry chef in Turkey, and a farmer in Kyrgyzstan. His tools included AI software built with models from American tech companies, allowing him to reach victims swiftly and on a massive scale.
Technology from the United States is being used to industrialize and globalize fraud, according to an AP and “FRONTLINE” investigation. Watchdogs argue these companies have the capacity to fight abuse but lack incentives. The Federal Trade Commission estimates scammers cost Americans nearly $200 billion in losses in 2024. Scams often start upstream of social media platforms. American technology supports every step in the digital supply chain, helping scammers evade internet restrictions and reach millions of victims.
The AP investigation did not find these companies committing illegal acts, but their technology was misused at scam centers in Myanmar. AI models like ChatGPT and Gemini help scammers work fluidly in multiple languages and monitor workers. Blockchain analysis by TRM Labs indicated these tools generated millions in illicit profits. Meanwhile, a robust global internet infrastructure supports scam operations in Myanmar, leveraging services from companies like Cogent Communications and AT&T. One in five signals from scam centers in Myanmar used U.S.-registered companies.
Starlink, operated by Elon Musk’s SpaceX, is the leading internet provider in Myanmar, serving scam centers despite pressure for a crackdown. Since a strict sweep last fall, at least 25 new scam centers have emerged in Myanmar.
Efforts to combat scams are underway. U.S. Attorney Jeanine Pirro formed the Scam Center Strike Force to disrupt scam networks. In a recent operation, it collaborated with Meta and Google, taking down over 1.4 million accounts linked to scams. OpenAI and Google say they have programs to prevent misuse of their tools, but challenges remain. Cybersecurity experts highlight that preventing scams requires resources, and currently, the cost for tech companies to facilitate scams is minimal.
The U.S. lags behind countries like the UK and Singapore in enforcing regulations against scams. However, American tech companies have started assisting in voluntary efforts to cut scammers from U.S. infrastructure. AI innovations used in scams outpace scrutiny, raising fears of fully automated scams. Koorimannil’s work required minimal human intervention, with scripts created by bosses that he just copied and pasted. The use of AI makes scam operations more efficient, sometimes replacing the need for human operators entirely.
Several large companies form a part of the digital pipeline that facilitates these scams. Internet service providers (ISPs) in the U.S. inadvertently support scams run from Myanmar by routing connections through their systems, a significant contribution to the problem. These ISPs claim they do not monitor user traffic directly but act on abuse reports. Some, like AT&T, have introduced new rules to improve transparency and prevent misuse.
Global initiatives to counteract scam activities have shown some results. In the U.K., U.S., and elsewhere, coordinated efforts have blocked scam infrastructure and pursued legal remedies. However, technological sophistication of scam operations continues to challenge these efforts.
The situation remains complex, and sophisticated internet tools enable scammers to evade detection while platforms struggle to keep up. Even as some companies improve monitoring and enforcement, scammers adapt and remain a step ahead. The commitment to tackle this issue requires wider collaboration among ISPs, tech companies, and law enforcement.