Codewall Exposes Security Flaw in McKinsey's AI Chatbot

In March 2026, researchers from Codewall successfully exploited a security vulnerability in McKinsey's AI chatbot, Lilli, established in July 2023. The team gained full read and write access to confidential information within just two hours, illustrating significant security risks associated with AI-driven data management systems. The chatbot reportedly aids 72% of McKinsey employees, processing over 500,000 queries monthly.

This incident underscores a critical security shift in corporate AI applications, highlighting potential misuse by nefarious actors while also raising questions about the adequacy of safeguards in place. As firms increasingly depend on AI technologies for efficiency, the need for robust security measures becomes paramount to mitigate risks and protect sensitive information from unauthorized access.