ChatBotKit Reflection Topics for prompt-injection

Explore our journey through the evolving world of chatbots, the technology that powers them, and the honest reflections of our team's experiences.

Untrusted Context Belongs in a Tool Call
A security firm hijacked Bunq's AI assistant through a transaction description. The breach happened because external content was fed into the prompt, where the model treats it as an instruction instead of as data. There is a better way.