Black market exploits AI models for cybercrime

According to a report by the Financial Times, AI models and computational power are becoming highly sought-after items in the dark web, with hackers using them for ransomware, cyber warfare, and espionage. John Hotteqvist, a senior analyst at Google's Threat Intelligence team, noted a significant increase in 'LLM hijacking' activities this year, including the sale of stolen login credentials for public AI tools and the misuse of others' computational resources. The dark web now offers access to AI models from companies like Anthropic, Google, and OpenAI at discounts up to 97%, compared to subscription prices of up to $200 per user per month. Hotteqvist warned that attackers can use these models at a fraction of the cost, giving them an economic advantage over defenders. He also highlighted the risk of cybercriminals implanting their own AI models into corporate cloud servers, a tactic similar to cryptocurrency mining on compromised devices. As more companies move AI infrastructure in-house, they become new targets for cyberattacks.