Google has filed a lawsuit against a Chinese cybercrime network, alleging the hackers are using the company’s Gemini artificial intelligence models and tools to build phishing software to rob consumers.
The lawsuit, filed Friday in the U.S. District Court for the Southern District of New York, alleges the network created a phishing software called “Outsider” that encourages hackers to use normal AI tools like Google’s Gemini to recreate more fraudulent websites.
The Outsider software allegedly tricks victims over text into thinking there is a problem with their brokerage accounts or that they are eligible for rewards through their phone carrier.
From there, a user is sent to a fake website where users are duped into giving their personal and financial information, the suit stated. Victims were allegedly tricked out of millions of dollars, according to Google.
The Outsider software has more than 290 prebuilt templates impersonating legitimate institutions like financial service providers, government agencies and retailers, according to the suit.
The group also made a tool with “step-by-step instructions on how Outsider can weaponize AI-generated code.” With the instructions, hackers can create the custom code needed to turn the templates into fraudulent sites.
“On their own, these prompts appear to be innocent requests for programming assistance,” the suit states, adding, “Using this method, Enterprise members can create convincing duplicates of virtually any legitimate website in minutes.
As a result, the 290 identified websites are “only a starting point,” calling the potential for more phishing sites “limitless.” The firm discovered more than a million URLs originating from the Outsider software.
The Hill was unable to reach an attorney for the 25 “Doe” defendants listed in the suit.
Google emphasized AI is enabling hackers to work at much faster speeds, stating it hopes the suit will disrupt the criminal enterprise to protect the online ecosystem.
The lawsuit comes amid growing concerns the role of AI in cybersecurity. These concerns seemed to reach a tipping point earlier this year after the AI firm Anthropic released its Mythos cybersecurity model to a select group of companies and government agencies.
Mythos, according to Anthropic, is the most advanced model to detect security vulnerabilities. This can be a tool for critical infrastructure to patch up vulnerabilities, but also a powerful weapon for bad actors targeting websites, software and other infrastructure.
The full story
This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.
