Meta's New AI Assistant Had a Flaw That Let Hackers Take Over

Meta has rushed out a fix for Muse, its brand-new AI assistant for Mac computers. A security researcher named Patrick Wardle found a flaw that let someone take control of the assistant — the kind of AI that can read your screen, take notes, and do tasks for you. Meta patched it within hours of the report becoming public. The company says the danger to normal users was low.

Here's the catch that made it scary. The flaw only worked if bad software was already running on your Mac — in other words, a hacker had to get in the door first. But once inside, they could quietly turn your friendly AI helper into a tool for spying. Wardle proved he could make Muse take pictures and write malicious files to your computer, often without any warning popping up. In plain terms: instead of building a complicated virus from scratch, an attacker could just borrow the powers you already gave your AI assistant.

Wardle's criticism was blunt: companies should build security in from day one, not bolt it on later. That stings because Meta promoted Muse's privacy and security features when it launched. The timing is awkward for other reasons too — Amazon recently blocked Muse from accessing its shopping site, saying Meta never asked permission. Yet the launch has been a hit: Muse's mobile app reportedly beat ChatGPT's first 12 days of downloads in the US and Canada, and Meta's stock jumped 11 percent on Monday.

What should you take from this? AI assistants are getting more access to your computer, your files, and your accounts. That's convenient — and it makes them a juicy target. If you use an AI helper on your desktop, keep your software updated, and think twice before handing it permission to do everything.