The fact
Researcher Patrick Wardle revealed that unprivileged software could exploit the assistant's extensive permissions.
Meta issued a temporary fix, but the vulnerability raises concerns about AI agent security.
Click the link to read an article on the topic: