Criminal syndicates in Southeast Asia are using large language models and deepfakes to expand online scam compounds and modern forced labor.
According to Atlas Diplomacy, an analytical report titled “How Artificial Intelligence Is Fueling Cyber Scams and Human Trafficking in Southeast Asia” by Rita S. Fernandez has been published in the Council on Foreign Relations. Drawing on field research and UN data, the report examines how organized crime networks in countries such as Cambodia, Laos, and Myanmar exploit artificial intelligence tools to traffic victims and steal tens of billions of dollars globally. Below is a summary of the piece.
The proliferation of cyber scam compounds across Southeast Asia—particularly in Cambodia, Laos, and Myanmar—has become a major driver of human trafficking and forced criminality. The International Organization for Migration (IOM) estimates that roughly 300,000 victims from at least 80 countries have been trafficked or coerced into working in regional scam centers, with IOM assisting more than 3,500 victims between 2022 and 2025. According to the UN Office on Drugs and Crime, financial losses from cyber scams in East Asia, Southeast Asia, Australia, and New Zealand reached between $88 billion and $114 billion in 2025, while victims in the United States lost $18.2 billion that year.
Artificial intelligence tools and satellite communications such as Starlink have significantly amplified the reach and efficiency of these criminal networks. Syndicates use AI to generate convincing job advertisements targeting university graduates and tech-savvy candidates, forge visas and corporate licenses, and deploy voice-cloning technology to lure victims into transit countries like Thailand before abducting them across borders into scam compounds.
Inside the compounds, victims are stripped of their passports, beaten, and forced to use large language models such as ChatGPT and Gemini to translate scam scripts into multiple languages, generate target lists, and operate specialized rooms using real-time face- and voice-manipulation tools to defraud victims worldwide.
Law enforcement agencies and technology firms have begun responding: the Royal Thai Police launched the SHIELD intelligence-sharing platform with partners including China, India, Japan, Australia, and the United States; Meta removed 1.4 million accounts linked to scam operations; and SpaceX disabled nearly 2,500 Starlink terminals connected to compounds in Myanmar. Still, the ease of stripping safety guardrails from open-source AI models and fragmented national regulations underscore the urgent need for a cohesive global AI governance framework./ Source
