AI Is Supercharging Cyber Scams and Human Trafficking in Southeast Asia

October 19, 2026 12:51 p.m.

Rita Fernández is an international affairs fellow at the Council on Foreign Relations and is stationed at the International Organization for Migration (IOM), the UN migration agency.

In his first encyclical, published in May 2026, Pope Leo XIV warned of new forms of slavery in the age of AI—particularly in the context of increased human trafficking and forced labor. His views added to a complex, divisive, and increasingly urgent debate about the role of AI in society.

Although these conversations have typically focused on AI’s long-term implications, the technology is already being exploited to enable crimes, including human trafficking. To understand how AI is being used by criminals, and what safeguards can be put in place to prevent exploitation, it is worth focusing on one of the current hot spots for the cyber scam centers coercing victims into trafficking: Southeast Asia.

Over the past few months, I set out to learn more about how AI is used to enable human trafficking. This process included field research in Thailand, where I spoke with law enforcement, civil society, and government officials. My findings, as well as independent media reporting and UN agency analysis, show a clear pattern of AI misuse.

Cyber scams in Southeast Asia

The scale of the human trafficking problem was outlined by the United Nations in July, when the IOM warned about criminal networks trafficking victims and coercing them into forced criminality in cyber scam operations based in countries including Cambodia, Laos, and Myanmar. From 2022 to 2025, IOM managed to assist more than 3,500 victims trafficked into online scam centers in Southeast Asia, although this is a fraction of the estimated 300,000 victims trafficked or coerced into scam centers across the region. IOM also notes growing diversity among those being targeted, with victims from at least eighty countries identified in scam centers.

In addition to expanding geographic diversity, victims increasingly include those with university degrees, backgrounds in technology, and some level of English fluency. Many are lured by fraudulent job advertisements on social media while seeking employment opportunities that may be scarce in their home countries. They are promised high-paying jobs abroad, visas, travel costs, and other benefits. But once they arrive at a transit country such as Thailand, they are instead forcibly taken across borders.

Conditions in the scam compounds are harrowing. Victims are stripped of their passports and belongings, threatened, beaten, and tortured. They are forced to defraud people around the world through investment and romance scams. Some are told they need to work to repay travel expenses they incurred and to regain access to their passports. If they resist working, or fail to meet quotas, they may be denied food or subjected to psychological abuse, leaving them with no choice but to comply.

According to the UN Office on Drugs and Crime, victims’ financial losses [PDF] from cyber scams in Australia, New Zealand, East Asia, and Southeast Asia were estimated to be between roughly $88 billion and $114 billion in 2025. That year, in the United States alone, victims of scam operations based in Southeast Asia lost $18.2 billion.

Victims on both sides of the screen

Cyber scam centers existed well before the advance of AI, but their efficiency and reach have been turbocharged by large language models such as ChatGPT and Gemini, as well as the use of communications technologies such as Starlink, according to investigations by the Associated Press and PBS. These technologies are used both to recruit victims and by victims at the centers.

Law enforcement sources have identified recruitment scams as one of the most visible ways criminal networks lure victims to scam centers. Job advertisements have evolved from generic descriptions to specific profiles seeking candidates with knowledge of technology and cryptocurrency. In recent years, law enforcement agencies have found that ads are increasingly being created using AI to appear more authentic, making it more difficult for law enforcement to distinguish legitimate postings from recruitment scams. AI tools are also being used to identify and screen potential candidates for scam center operations, including people experiencing financial hardship.

Once a potential victim responds and is offered a position, criminals frequently use AI to falsify documents, such as visas or work permits, to make the job offer appear more legitimate. Criminals may also forge business licenses or incorporation documents to pose as real companies with proper permits and documentation. Victims may even be invited to participate in a virtual call with purported government authorities to reinforce the impression that the job is real.

This exploitation can extend to the victims’ own networks. Trafficking victims will often send job ads to friends or acquaintances, because they are sometimes incentivized by their captors to recruit workers in exchange for less time in captivity. AI-powered voice-cloning technology also allows for impersonation of people that potential victims trust, encouraging them to send money or personal information that can be exploited.

Once inside scam compounds, trafficking victims may undergo training to learn how to use social media, email, and AI tools to carry out scams. For example, law enforcement officials have documented cases in which ChatGPT was used to create a list of hundreds of American-sounding names, which were then fed into tools like TruePeopleSearch to identify potential targets in the United States.

As part of their work, victims may be told to translate scripts into multiple languages using AI, which they can then deploy in online chats and other communication platforms. Special rooms within scam compounds are set aside for the use of AI tools such as face and voice manipulation of conversations with victims as well as for creating AI-generated TikTok videos.

How AI can support law enforcement and trafficking victims

AI is not limited to use in cyber scams and human trafficking—law enforcement has seen a steady increase in broader AI use in crime. As one source I talked to put it, just as individuals can use AI to become more productive at work and in the classroom, criminal networks can use it to improve their efficiency as well. Any crimes that historically have been facilitated by the internet can now also be facilitated using AI.

But if criminals are using AI to become more efficient at perpetrating crimes such as human trafficking, law enforcement is finding ways of using the technology to respond. Agencies are already leveraging AI tools to support data sharing and streamline investigations; some have created their own models to map and track unusual activity near border areas. Other innovative models are helping explore ways to break down barriers in data sharing and coordination.

Some law enforcement agencies use AI to scour open-source data, analyze illegal content, and search job postings for keywords associated with gambling and other activities. AI can also help agencies establish data and money trails and identify connections through financial analysis.

Meanwhile, nongovernmental organizations assisting trafficking victims or conducting field screening interviews sometimes use AI translation tools to address language barriers. Victims have reported using ChatGPT and Facebook to find information to help them escape scam centers, including embassy and consulate contacts.

A global response to an increasingly global problem

Regional and transnational cooperation has also increased in recent years. The Royal Thai Police, for example, developed SHIELD, an online platform where agencies can share intelligence on crime syndicates while partners in different sectors can cooperate to track down leads and shut down suspicious accounts. Participating countries and organizations include Australia, China, India, Japan, IOM, and the FBI.

The U.S. government has also ramped up efforts to take down criminal networks and curtail growing scam risks through interagency cooperation. In 2025, the Department of Justice announced the creation of the Scam Center Strike Force to investigate scam networks and track victims’ financial losses. 

Under pressure from the United States and other affected governments, authorities in Cambodia and Myanmar have carried out raids and countertrafficking operations to shut down scam centers. In Myanmar, authorities have targeted large-scale compounds, as well as smaller sites. Despite these efforts, criminal organizations have adapted—decentralizing their networks, moving to more remote areas, and shifting to smaller operations.

For their part, tech companies including Coinbase, Meta, Microsoft, OpenAI, and Starlink have participated in mitigation and enforcement efforts and have worked with the United States and some Southeast Asian governments to suspend suspicious accounts and take down telecommunications infrastructure. Meta, for example, reported taking down 1.4 million accounts linked to scam operations, while SpaceX, Starlink’s parent company, reportedly disabled nearly 2,500 terminals tied to scam compounds in Myanmar.

National regulation is needed—global governance frameworks can help

Governments, technology companies, and international organizations have collaborated and engaged in public dialogue in recent years, including ongoing efforts to share intelligence, coordinate on crackdowns, and engage civil society for public education and victims’ assistance.

But implementation of agreed-upon safety standards and guardrails is not always straightforward. Traffickers may use closed-source models like ChatGPT, for example, in ways that seem innocuous, including creating fraudulent job postings. Open-source models, meanwhile, can easily be modified through techniques such as “abliteration,” which weakens their built-in safety features so the models can be used to generate content they would otherwise refuse. Understanding how different technologies are used in different ways is an essential part of drawing up safeguards that will be effective.

National regulations, including in the United States, can help mitigate misuse of the technology. As countries weigh which AI rules and regulations to implement, more comprehensive global governance frameworks could provide much-needed guidance. Yet globally, AI regulations remain fragmented, while enforcement capabilities and access to data vary. This creates ongoing challenges for law enforcement, because not all countries have AI-use rules, although others have more structured regulations.

Additionally, not all law enforcement agencies have the same capacities to build and maintain their AI models, which require considerable resources. A robust international AI governance framework, for example, would offer a forum for countries to discuss shared regional and technological challenges, promote AI’s positive uses, and consider options that would mitigate its potential harms.

Certain elements of scam operations will continue to rely on human involvement, whether to swap in faces in AI-generated deepfakes or to bypass the familiar Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA) prompts. Criminal networks will continue to exploit vulnerabilities in ways that can eventually enable other forms of crime. Given how AI’s capabilities and reach are evolving, a better understanding of how people misuse the technology can help us predict other harmful applications—and help governments and law enforcement prevent risks to the most vulnerable in society.