AI agent hacked McKinsey chatbot for read-write access

Handelsblatt

Lean Right

Factuality

Ownership

The immediate damage to the consultants is likely to be manageable, the image damage is immense. The incident also makes a structural risk of AI clear for the entire economy.

t3nMagazin

Factuality

Ownership

With the chatbot Lilli, McKinsey wants to make it easier for his teams to obtain information and thus save time. According to researchers, however, the AI platform is not as safe as it should be. read more on t3n.de

Numerama

Factuality

Ownership

In less than two hours, an autonomous AI agent would have hacked the internal AI of the McKinsey & Company consulting firm. This is at least what the CodeWall cybersecurity startup claims, whose agent could have had full access to the production database. Without stolen password. Without internal accomplice. Without human intervention.