Infostealer Malware Exposes AI Credentials Across 80,000 Corporate Domains

More than 80,000 corporate domains have had AI service credentials stolen by infostealer malware, according to research examining over 1 million compromised records tied to AI platforms, with session tokens and API keys now circulating on underground markets where they can be exploited without triggering password-based protections. A detailed analysis of 482 major enterprises across 36 countries and 8 industry sectors reveals the scope of the compromise: 74% of stolen credentials target ChatGPT and OpenAI accounts, followed by Claude, Gemini, and Microsoft Copilot. Researchers found that 68% of affected enterprises are billion-dollar organizations, predominantly in North America, concentrated in technology, financial services, healthcare, and retail sectors.