An AP/FRONTLINE investigation reveals global scammers are exploiting American AI tools and US-based infrastructure to run scalable romance and other schemes from Myanmar, targeting victims worldwide in dozens of languages.

Scammers rely on platforms like Kongtian Intelligent Customer Acquisition (KT) and Global Social Traffic Navigation (007TG), leveraging AI models from OpenAI, Google Gemini, and others to automate replies, translate, and monitor workers for high-speed, cross-language fraud.

The use of AI tools—especially OpenAI’s ChatGPT and Google’s Gemini—enables Southeast Asian operations to automate multilingual scams, create personas, monitor performance, and generate tens of millions in illicit revenue.

There is a tension between privacy-by-design and the ability to monitor abuse, as providers say they cannot see content or end-user activity but will respond to abuse reports and cooperate with law enforcement.

Privacy-by-design limitations limit end-to-end monitoring, restricting detection or blocking of illicit activity without user reports and law enforcement cooperation.

Starlink is highlighted as a major provider for scam centers in Myanmar, prompting policy questions about curb measures while Starlink argues it serves legitimate users; ongoing regulatory actions and investigations are in progress.

Public watchdogs and policymakers in the UK, EU, Australia, and Singapore are imposing or considering penalties to enforce stronger anti-scam measures, while U.S. authorities seek voluntary cooperation from American tech companies to curb scam activity.

U.S. policymakers urge voluntary cooperation from tech firms to cut off scammers from US infrastructure, with officials calling for industry readiness to stop fraud when detected.

Regulatory landscape is evolving: anti-scam regulations have been introduced in multiple regions, and the U.S. is pursuing voluntary cooperation from tech platforms to shut down scam networks from infrastructure.

Law enforcement and industry responses include the DC Scam Center Strike Force disrupting millions of accounts and infrastructure, with actions from OpenAI and Google to limit scam use of their tools; Starlink did not comment on specific requests.

The investigation notes that while it does not prove illegal activity by firms, abuse patterns raise questions about terms-of-service enforcement and the effectiveness of self-regulation across the tech ecosystem.

The investigation estimates nearly $200 billion in American scam losses in 2024, underscoring the financial stakes and the need for stronger enforcement and incentives to prevent abuse.