Meta's new personal assistant, Muse, is not just your everyday chatbot. It's meticulously archiving your social life, creating detailed profiles of everyone in your life, from family to colleagues. These profiles include intimate details like dates and shared experiences, offering a level of personalization that might make even the creepiest stalker blush.

Independent researcher Karan Joshi managed to extract these internal instructions, revealing a system that compiles data on your relationships, suggesting ways to strengthen them. This might seem harmless, but given Meta's vast access to social network data, it's a concerning step towards a digital shadow of your real life.

Meta defends this feature, stating that Muse is designed to seek human confirmation before making significant actions and keeping a record of all its activities. However, the sheer amount of data it collects raises questions about privacy and consent. As AI assistants become more integrated into our daily lives, the line between helpful and invasive is becoming increasingly blurred.

Professor Carissa Véliz warns that we're giving AI systems much more information than we’re getting back, with potential for both correct and incorrect inferences. The future of AI lies in how we balance utility with privacy, a challenge that Muse is certainly pushing to the forefront.