Google Report: Dark Web AI Model Access Sold at Up to 97% Off

John Hultquist, chief analyst at Google's Threat Intelligence team, revealed that illegal access to AI models and computing power is rapidly becoming one of the hottest commodities on the cybercrime black market, with so-called "LLM hijacking" activity increasing notably this year.

Dark web markets are selling access to AI models from companies like Anthropic, Google, and OpenAI at discounts of up to 97%. For comparison, the highest-tier subscriptions for ChatGPT and Claude can cost up to $200 per user per month. Some sellers even offer "guaranteed access" services, promising free new credentials if an account gets banned.

Hultquist warned that attackers can use otherwise expensive AI at low prices, giving them a cost advantage. Some criminal groups and state-backed organizations break into corporate cloud servers and plant their own AI models, using victims' systems to run them.

As more large enterprises deploy custom AI models on their own servers, these self-run AI systems will also become attack targets. Hultquist believes now is the best time for hackers to "sneak into" target accounts and computer servers.