Dark Web Marketplaces Selling Hijacked AI Model Access at 97% Discount Through LLM-Jacking Attacks

Google Threat researchers discovered that dark web marketplaces are selling access to AI models from companies including Anthropic, Google, and OpenAI at discounts reaching up to 97%. This represents a new attack vector called LLM-jacking, which involves hijacking or cryptojacking of Large Language Models. The attacks allow threat actors to hijack corporate accounts and use them to target third parties at minimal cost, creating a new form of economic asymmetry that favors attackers.