This sneaky photo trick gets AI chatbots to ignore their safety rules

This sneaky photo trick gets AI chatbots to ignore their safety rules

💻 Technology & Science
Summary

A new exploit out of Florida International University shows how invisible pixel-level changes in a photo can trick AI chatbots into generating responses they would normally block.

Visit the original source for the full article

🌐 Translate to:
More from 💻 Technology & Science →