Google sues Chinese cybercrime ring that used Gemini to build phishing sites and send 2.5 million scam texts

Google filed a lawsuit on Friday to dismantle the infrastructure behind a Chinese cybercrime operation called Outsider Enterprise. The group used AI, including Google’s own Gemini, to generate phishing websites and send scam text messages impersonating Google and other brands. It sent 2.5 million fraudulent texts to Android users in a two-week period.

The operation deployed 9,000 fake websites and 1 million fraudulent web domains designed to steal passwords and credit card numbers. Google said the group has financially scammed “hundreds of thousands of victims” with losses “estimated in the millions.” In just two weeks in May, 55,000 spam texts were flagged by Android users, more than two complaints per minute.

The most damaging detail is in the court filing. Members of Outsider Enterprise actively encouraged each other to use Gemini to generate custom code for phishing websites, which was then imported into the group’s software suite and converted into live scam pages. Google’s own AI was used to build the tools targeting Google’s own users.

The group coordinated through Telegram and distributed “phishing kits” that allowed lower-level criminals to launch fake text campaigns mimicking trusted brands. Google said it uses “AI-powered tools to fight AI-powered scams,” intercepting more than 10 billion scam messages per month through its detection systems.

Google said it is coordinating with the FBI, which is taking unspecified law enforcement actions. The company is also working with AT&T, T-Mobile, and Verizon to block the scam texts before they reach users. The FBI did not respond to a request for comment.

The lawsuit is civil, not criminal, meaning Google is seeking to shut down the infrastructure rather than put anyone in prison. It is the kind of action Big Tech companies increasingly use to go after cybercrime operations when law enforcement moves slowly or not at all, particularly when the perpetrators are in jurisdictions beyond the reach of Western authorities.

The scale of the operation underscores how cheap AI-powered scams have become. Building thousands of convincing phishing pages used to require skilled developers. Now a Telegram group can use a frontier AI model to generate the code and deploy it at scale. AI is making attack tools cheaper across every category, from vulnerability discovery to social engineering. Outsider Enterprise is what that looks like on the distribution side: industrial-scale fraud built with off-the-shelf AI.

Google said it will continue to invest in AI-powered scam detection and urged users to enable its spam protection features on Android. But the uncomfortable truth is that the same company selling the AI is now suing the people who used it for crime. AI agent security is not just a product feature problem. When the tools are this powerful and this accessible, the arms race between builders and abusers is permanent.