AI agent breached McKinsey's internal platform in two hours

Security firm Codewall deployed an autonomous AI agent against McKinsey's internal AI platform, Lilli, with no credentials, no insider knowledge, and no human assistance. Within two hours, the agent had gained full read and write access to the production database. The attack relied not on novel exploits but on a decades-old technique, raising immediate questions about how enterprise organizations are securing AI systems deployed at scale.

Lilli serves more than 43,000 McKinsey employees across strategy work, client research, and document analysis -- making this a high-impact breach scenario. The demonstration shows that AI agents can now automate the reconnaissance and exploitation steps that previously required a skilled human attacker working over days or weeks. Organizations deploying internal AI platforms may be exposing themselves to attack vectors they have not yet mapped, particularly if security reviews were designed for conventional software and not adversarial AI agents.