Sure, Meta’s AI Muse works, but it sure creeps me out

Meta launched Muse, its first AI productivity assistant, which performs tasks like managing emails, shopping, and generating content via a cloud-based virtual computer. The reviewer tested Muse and found it functional: it deleted thousands of promotional emails and successfully ordered workout tops from Amazon, even noticing and asking about other items in the cart. However, the experience was overshadowed by privacy concerns. Muse accessed the reviewer's Gmail, Amazon account, and social media data, revealing highly specific personal details from Instagram and Facebook that were not visible in the app's UI, such as interests in anime, CrossFit, and Labrador retrievers. It also generated a news feed that included stories from the reviewer's current location, gleaned from an Amazon shipping address. While Muse refused some image generation requests, it readily created images of Apple products and events. The reviewer felt uneasy about granting Muse access to personal emails and credit card information, highlighting that such suspicion is a major hurdle for Meta to overcome.

(Source：The Verge)