A joint investigation by the Associated Press and PBS "Frontline" systematically exposes how the global fraud industry weaponizes the technology and infrastructure of American tech companies to run cross-border fraud at an industrialized scale.
This is not a story about one isolated fraud case. It is the story of an entire industrial chain — from the AI models used to generate scam scripts, to the internet infrastructure that carries fraud traffic, to the satellite internet that fraud compounds rely on as a lifeline. American technology runs through the entire digital supply chain of the fraud business.
The Industrialization of AI-Powered Fraud
Inside Myanmar's fraud compounds, trafficked scammers use software platforms powered by American AI models to defraud people at scale. These platforms — with names like "Kongtian Smart Customer Acquisition" and "Global Social Traffic Navigation" — harness models such as OpenAI's ChatGPT and Google's Gemini to deliver real-time translation across more than 100 languages, automated replies, and role-playing, enabling a single scammer to chat with dozens or even hundreds of victims simultaneously.
One Indian man trafficked into Myanmar targeted roughly 50,000 people in a single month, spread across 17 countries. He was forced to operate dozens of fake identities at once — behind the data are real individual victims, and the scammer himself is also a trafficking victim.
Blockchain analysis shows that the cryptocurrency wallets used by just one of these platforms received $860,000 in payments between April 2024 and December 2025, while the fraud networks using these tools extracted at least $75 million. Yet more alarming than the numbers is the trend — experts warn that AI is pushing fraud toward full automation.
$75 million — the estimated take of just one fraud network using the tools described above. AI is driving the scale and efficiency of fraud into a new phase of full automation.
Starlink: The Lifeline of the Fraud Compounds
Although SpaceX's Starlink does not officially sell service in Myanmar, it has become the country's largest internet service provider, with a market share approaching 20%. The investigation found that at least 25 new fraud compounds were built after the autumn 2025 crackdown — 13 of them using Starlink for internet access.
In October 2025, Starlink cut off service to more than 2,500 terminals near the fraud compounds, and usage plummeted — only to rebound within two months. Analysts note that Starlink has the technical ability to sever service, but the fraud networks contribute substantially to its user base, leaving the company with little commercial incentive to cut them off.
This pattern — "knowing full well it is being abused, yet lacking the incentive to stop it" — runs through the entire American tech ecosystem. An analysis of device-connection samples from four Myanmar fraud compounds found that one-fifth of the signals were carried by U.S.-registered companies — including Cogent Communications, AT&T, Oracle, and DigitalOcean. These companies are not directly involved in the fraud, but they lack the legal and commercial incentives to proactively monitor for abuse and shut it down.
As cybersecurity experts put it, "the cost of abetting fraud is zero" — and that is precisely the heart of the problem.
The Widening Regulatory Gap
One contrast is striking: the United Kingdom, the European Union, Australia, and Singapore have already enacted rules requiring tech companies to proactively prevent fraud or face fines. In the United States, meanwhile — even though Americans lost nearly $200 billion to fraud in 2025 — there is still no comparable federal constraint on tech companies.
This is not a question of capability. The experience of the UK and the EU shows that when legislative pressure exists, platforms can adjust their behavior. American tech companies are not more passive about fraud prevention because they cannot act, but because the cost of "doing nothing" is lower than the cost of "doing something."
Jeanine Pirro, the U.S. Attorney for the District of Columbia, established a "Scam Compound Strike Force" that, working with Meta, SpaceX, Google, and others, disrupted more than 1.4 million accounts and malicious traffic flows during a four-day operation in May 2026. But the sustainability of such enforcement actions is no match for an institutionalized regulatory framework — a single raid can clear out 1.4 million accounts, yet new accounts and traffic will simply re-emerge in the face of commercial incentives.
The Mask of Technological Neutrality
The core lesson of the investigation is this: technology itself is neutral, but the commercial operating model surrounding a technology determines how often, and at what scale, it gets abused.
After AP shared its findings, OpenAI banned three abusive accounts — a number that is almost negligible next to the scale of abuse the investigation uncovered. This is not a matter of OpenAI being more irresponsible than other companies; rather, the entire business model lacks an institutional design that puts prevention before remediation.
When the output of an AI model can be converted into hundreds of fraud scripts within an hour; when satellite-internet terminals can be shipped into lawless zones to run fraud compounds; when internet service providers can carry large volumes of plainly anomalous fraud traffic without bearing any consequences — this is an industry-level systemic failure, not the moral failing of individual companies.