How the romance scam industry actually works — and who is actually in the compounds

The standard expat conversation about romance scams goes something like this: foreign man meets Thai woman, foreign man loses money, everyone nods knowingly.

Almost every part of that framing is wrong.

An online profile does not reliably identify the person operating it. The United Nations and UNODC have documented trafficking into scam operations across the region, including people recruited through false job offers and forced into criminality. They also distinguish coerced workers from managers and organisers in transnational criminal networks.

Thailand’s survey and complaint data show a serious domestic fraud problem. Around six in ten Thai adults report having been scammed in one survey. That is a finding about Thai respondents, not a nationality breakdown of everyone who loses money in Thailand. No published nationality breakdown of Thai scam losses was located for this article.

Why this matters

Because the mythology is doing real damage.

It teaches foreigners to infer criminality from a person’s nationality or profile, which is both insulting and a poor defence. It also obscures an industry in which UN reporting documents trafficking, forced criminality and grave abuses alongside organised criminal management.

Thailand is geographically and institutionally connected to the regional problem, but this article does not claim that Bangkok is the general recruitment gateway for every compound. The documented routes, actors and locations differ by case.

The scale, which is difficult to take in

A wicked problem, published by the UN Office of the High Commissioner for Human Rights in February 2026, puts it this way: the industry has grown to “industrial proportions”, with credible estimates pointing to at least 300,000 people working in scam operations across Southeast Asia. Victims came from 66 countries as of March 2025.

One number in that report is regularly misquoted, including in the draft of this article. The report says 74 per cent of scam compounds are in the Mekong region, established from satellite imagery. That is a count of sites, not of people. It does not tell you that three-quarters of the workforce is there, and this article previously said it did.

Separate, earlier UN assessments put around 220,000 people in forced scam labour across Myanmar and Cambodia specifically.

These come from different exercises with different methods and scopes. They do not nest inside one another and must not be added together. Estimates in this sector are necessarily rough; nobody is running a clean census of a criminal industry, and “at least” is doing real work in every one of these figures.

The financial estimates in this sector are difficult to compare. They use different definitions, surveys, reporting systems and date ranges. This article has withdrawn its previous Cambodia revenue-to-GDP comparison: an estimated illicit revenue number is not a measured share of a country’s national accounts.

In Thailand, the Global Anti-Scam Alliance’s 2025 survey estimates losses of ฿115.3 billion a year from a sample of 1,000 adults. That is an extrapolated survey estimate, not a police count. Reported police complaints are a different measure. Neither source supplies the nationality split that the old article implied.

The practical point does not require a flashy total. Romance scams use sustained deception and psychological pressure. A victim’s repeated transfers are evidence of a designed fraud, not evidence that they were foolish.

Who is actually sending the messages

Here is the part that inverts the whole conversation.

UNODC describes recruitment of young, educated and multilingual people for apparently legitimate jobs, followed in documented cases by confinement and coercion into online fraud. UN human-rights reporting describes grave abuses against people trafficked into these operations.

That evidence matters when we talk about scam work. The profile on a victim’s screen tells us neither who is typing nor whether that person is coerced. It is possible for an operation to include trafficked workers, willing participants and organisers at the same time.

That is why “be careful of Thai women” fails as advice. It substitutes a stereotype for the only useful warning: never treat a romantic contact’s request for money or investment as proof of a real relationship.

Who is actually losing the money

The English-language conversation often assumes a Western foreigner is the typical mark. The available Thai data does not answer that question by nationality.

Around six in ten Thai adults report having been scammed in the Global Anti-Scam Alliance survey, and around seven in ten report encountering an attempt. That tells us fraud reaches a large share of Thai respondents. It does not establish who loses the most money across every nationality, whether every case is reported, or who operated each fraud.

Foreigners are targeted and can lose life-changing sums. Thai residents are also heavily targeted. The defensible conclusion is not a demographic contest; it is that scams are transnational and a profile, language or nationality is not a reliable guide to who is behind the fraud.

Why the compounds are where they are

Look at a map of the operations and the pattern is immediate: they cluster in places where the Thai state stops and something less organised begins.

Myawaddy, on the Myanmar side opposite Mae Sot. Border zones in Laos. Sihanoukville and the Cambodian coast. These are territories with weak or contested governance, cheap electricity and fibre run in from neighbouring countries, and, critically, proximity to Bangkok’s airport.

Thai authorities have a role in prevention, victim protection and cross-border cooperation. This article does not generalise an individual recruitment route into a claim about Suvarnabhumi or Bangkok as a whole.

Common misconceptions

“It’s Thai women scamming foreign men.” Do not infer the operator’s identity from the profile. Regional operations are transnational and documented scam compounds use both criminal organisers and trafficked labour.

“You’d have to be an idiot to fall for it.” The 42-transfer case took 18 months. These are long-form operations run by people trained in exactly this, working to quotas, with scripts refined across thousands of attempts. Contempt for victims is one of the reasons the crime works — it stops people reporting.

“It’s a crypto thing.” Crypto is the laundering layer, not the crime. The crime is a relationship, built patiently over months.

“The Thai police don’t care.” Enforcement and victim protection involve difficult cross-border questions. Thailand and other states have taken actions, but raids and announcements are not a measure of every agency’s effectiveness or of a victim’s chance of recovery.

“If they shut the compounds, it ends.” It relocates. Analysts tracking the sector describe an industry that moves rather than closes — pressure in one border zone produces expansion in another.

What’s changing

UN reporting warns that enforcement action can be followed by relocation or rapid replacement of workers. It does not justify a confident forecast about any specific compound, country or policy. Treat current enforcement news as a time-sensitive update, not proof that the underlying abuse has ended.

Final thoughts

There’s a version of this story that’s comfortable for foreigners in Thailand, and it’s the version that gets told: be careful, they’ll take your money, look at that couple.

The actual story is bigger and worse than a stereotype about one nationality or gender. Regional organised-crime networks have used trafficking and coercion; people who lose money can come from many countries; and the available data has material limits.

None of which means abandoning caution with money. It means the caution should target the request, not the profile: never send money or valuables to an online romantic contact you have not independently verified, and treat an investment suggestion from that contact as a serious fraud warning. Time and flattery are not proof of authenticity.