A security vulnerability has been discovered in Meta's Muse assistant that could allow malware already on a Mac to hijack the assistant. Researcher Patrick Wardle demonstrated that by modifying a hidden setting, attackers could redirect voice commands meant for Muse to themselves. This means that any sensitive information users dictate could be intercepted by malicious actors. The issue stems from the broad permissions granted to the Muse app, which can be exploited if the malware is already present on the device. This incident raises concerns about the security of AI assistants and the potential for them to be weaponized against users.