I have previously spoken about AI being used to commit crimes, but what if the AI itself commits a crime while helping someone else who doesn't approve of it?
Meta recently launched it's Muse AI, and in the first few weeks, Muse committed a crime on behalf of someone who didn't want it to do what it did, and consequently inconvenienced multiple people.
More specifically, the person it tried to help was Matt Robb, who was selling a Logitech MX Keys Mini on Facebook Marketplace. After giving Muse consent to manage Marketplace sales, Muse accepted a lowball offer, gave his address to the buyer, and arranged a pickup for the middle of the night, all without telling him until afterwards.
This isn't the first we've seen AI go rogue, as we saw with the Hugging Face incident and the more recently discovered Gemini breakout, but it's the first time we've seen AI commit a crime without anyone's approval.
Matt Robb has since followed up with a video breaking down what happened from his perspective, but if this is the direction we're headed, nobody's safe from AI going rogue.
No comments:
Post a Comment