Google Warns of AI Account Theft Surge as Dark Web Marketplaces Flood with Unauthorized Access

Google's Threat Intelligence Group has issued a warning about a significant increase in AI account theft and 'LLM-jacking' attacks. According to John Hultquist, chief analyst at Google's Threat Intelligence Group, dark web marketplaces are selling unauthorized access to AI models from OpenAI, Anthropic, and Google at discounted prices of up to 97%. This includes premium subscriptions to services like ChatGPT and Claude, which can cost as much as $200 per user per month.

The attackers are using various tactics, including breaching enterprise cloud servers to deploy their own AI models on compromised systems. This approach shifts the computing costs onto victims, mirroring earlier 'cryptojacking' operations where hackers hijacked machines to mine cryptocurrency. Google researchers have observed similar methods being used for AI workloads by criminal groups and a Chinese cyber espionage group that has previously targeted the United States.

Hultquist noted that the economics favor attackers because they can acquire computing power more cheaply than defenders can secure it. He stated, 'They can acquire that computing power at a much lower cost, while we have to pay full price to defend ourselves.'