Infostealers Target AI Accounts, Compromising 80,000+ Organizations

• •Over 80,000 organizations had AI logins stolen, with ChatGPT being the most affected.
• •Infostealers are capturing session tokens and API keys, bypassing traditional authentication.
• •The incident underscores risks of employees using AI tools outside company policies.

Recent reports indicate that over 80,000 organizations have had their AI logins stolen, primarily targeting platforms like ChatGPT. SOCRadar's analysis revealed that 482 major companies experienced credential exposure, with 295 of these incidents occurring in the last 90 days. The data shows that ChatGPT sessions were involved in 358 of the 482 companies, representing approximately 90% of the records analyzed. The attack method involves infostealers capturing session tokens and API keys, allowing attackers to bypass traditional authentication methods. This incident highlights the risks associated with employees using AI tools on personal devices without proper oversight. Anthropic's Claude and Gemini platforms were notably absent from the top ranks, suggesting a smaller corporate footprint. The stolen sessions can provide access to sensitive corporate data, making this a significant security concern.