A €0.01 transfer could compromise bunq’s AI financial assistant.
On June 5, 2026, security researchers at Blue41 discovered that sending a one‑cent Euro payment to a bunq account could trigger the bank’s AI agent to execute arbitrary commands. The flaw stemmed from the assistant’s parsing of transaction metadata, which failed to sanitise user‑supplied strings before feeding them to the language model. Blue41 reported the issue to bunq, and the bank rolled out a fix within 48 hours, updating the input validation logic and revoking the vulnerable endpoint.
The bug matters because it showed that even minuscule financial actions can become attack vectors against AI‑driven services. Fintech firms increasingly rely on conversational agents for customer service and transaction handling; a similar oversight elsewhere could let attackers exfiltrate data or initiate unauthorized transfers.
Bunq’s quick response limits exposure, but the episode underscores the need for rigorous security reviews of AI integration points, especially when financial triggers are involved.