Research explained
Lilli security report describes conventional application flaws reaching an AI knowledge base
Codewall’s reported research involved exposed endpoints and SQL injection; its scale and speed claims remain attributed to the researcher.
What happened
The Stack reported that Codewall researchers accessed data associated with McKinsey’s Lilli assistant through exposed endpoints and a SQL-injection weakness. Codewall said its AI-assisted testing reached substantial chat and document metadata.
The reported record counts, time and cost are researcher claims, not independent BugSkan measurements. The access path described is an application-security failure around an AI system, rather than a vulnerability in every underlying model provider.
The report concerns a disclosed research finding, not established criminal theft of the entire dataset. Organizations with retrieval assistants should test API authentication, query construction and authorization across both source documents and conversation records.
Next step
Test API authentication and record-level authorization around retrieval assistants.
What remains unknown
- Open question
- The sources reviewed do not establish whether your deployment was exploited.
Source references
-
THE