AI Accounts Costing $200 a Month Are Being Sold for Just a Few Dollars

Companies' expensive artificial intelligence (AI) model accounts are being stolen by hackers. John Hultquist, principal analyst at Google's Threat Analysis Group (TAG), told the Financial Times (FT) on the 27th (local time) that there has been a sharp increase this year in cases of hackers illicitly seizing companies' high-cost AI model accounts and cloud computing resources (computer processing power rented over the internet).

Investigations found that subscriptions to major AI companies such as Anthropic, Google and OpenAI are being listed on the dark web at discounts of up to 97%. AI companies are responding by monitoring signs of abuse and blocking problematic accounts. However, dark web sellers are keeping the trade going by offering a "connectivity guarantee"—promising to provide new login credentials at no extra charge if the original account gets blocked.

Analyst Hultquist stressed that now that companies have begun deploying their own AI infrastructure, spikes in computing usage attract less suspicion, making it an opportune moment for hackers to secretly infiltrate targets' servers. The analyst issuing this warning is a security expert who in 2014 exposed the identity of the Russian hacking group 'Sandworm' and, two years later, proved the group was behind attacks on Ukraine's power grid.