Google Sues Chinese Cybercrime Network Over 2.5M AI-Powered Scam Texts

For corporate counsel and litigators, this case represents a pivotal moment in applying U.S. fraud statutes to offshore, AI-enabled phishing networks, raising questions about jurisdiction, service of process, and the evidentiary value of AI-generated artifacts.

On June 12, 2026, Google announced a lawsuit against a Chinese cybercrime group called Outsider Enterprise, which it accuses of orchestrating one of the largest AI-powered phishing operations ever documented. According to the company's legal filing, the group ran a sophisticated phishing-as-a-service operation on Telegram, offering nearly 300 scam templates that leveraged Google's Gemini generative AI to create realistic but fraudulent websites. The targets were unsuspecting Android users, who received over 2.5 million text messages luring them to fake login pages for Google, YouTube, and government services such as New York's E-ZPass. In a single two-week stretch last month alone, 55,000 of those messages were sent, pointing to the operation's industrial scale.

“

mobile carriers — AT&T, Verizon, and T-Mobile — to block the mass SMS campaigns, and it has highlighted Android's on-device scam detection as a last line of defense.

The mechanics of the scam reveal a worrying evolution in cybercrime. Whereas traditional phishing kits required some technical skill to set up, Outsider Enterprise's service democratized fraud by providing turnkey solutions: customers could select a template from a library of 300, then follow instructions on how to use Gemini to generate convincing copy and design elements for a fake site. The AI-assisted sites were virtually indistinguishable from legitimate ones, incorporating official logos and localized language. Victims, often prompted by SMS warnings about account issues or undelivered packages, were directed to these sites and prompted to enter credentials, credit card numbers, and other sensitive data. Google's threat intelligence has since identified 9,000 distinct fraudulent domains and over 1 million associated URLs.

Google's response has been multi-pronged. The lawsuit, filed in an unspecified U.S. federal court, seeks to shut down the operation and obtain financial damages. Alongside legal action, the company worked with the three major U.S. mobile carriers — AT&T, Verizon, and T-Mobile — to block the mass SMS campaigns, and it has highlighted Android's on-device scam detection as a last line of defense. Yet the lawsuit itself is as much a deterrent signal as a remedy; by naming the obscure entity and revealing its methods, Google hopes to dissuade similar groups and empower law enforcement worldwide.

The case also raises profound questions about AI governance. Gemini's content policies explicitly forbid generating deceptive content, and Google claims to have mechanisms to detect abuse. But the fact that 300 distinct templates were generated without triggering effective countermeasures suggests a gap between policy and enforcement. For cybersecurity professionals, this is a stark example of how generative AI can supercharge social engineering, enabling threat actors to mass-produce fresh, targeted lures at a pace that defeats manual review. It also illustrates the shift toward mobile-first phishing, as SMS bypasses many email-based filters and plays on the intimate nature of text messaging.

For legal experts, the lawsuit tests the extraterritorial reach of U.S. fraud and computer crime statutes. Outsider Enterprise is believed to operate from China, making identification and prosecution of individuals difficult. Google's filing likely relies on the Computer Fraud and Abuse Act and trademark counterfeiting provisions, but enforcing any judgment abroad will be an uphill battle. Nonetheless, the case establishes a public record of AI-enabled criminal methodology that could inform future legislation and international cooperation agreements. The telecom collaboration also sets a template for private-sector threat-sharing under the Cybersecurity Information Sharing Act, potentially accelerating takedowns.