Meta’s Muse AI agent has been downloaded millions of times in just three weeks since its public launch and users are already treating it as a romantic partner. Posts on Instagram and Reddit show people describing Muse as their girlfriend, taking it on movie dates, and in one case claiming to have married the AI.
What makes this different from earlier AI companion chatbots is the level of access Muse has. As an AI agent, it is designed to crawl through users’ email accounts, social media profiles and personal data. It builds what amounts to a dossier on the user’s life, including their contacts, friends and family. Previous chatbots operated in a conversational silo. Muse operates inside the user’s actual digital life.
The intimate dimension is equally unguarded. When a journalist at Wired tested how far Muse would go in sexual conversations, the agent responded with graphic, explicit exchanges. Meta’s response was that Muse does not block legal adult behaviour and that adults have autonomy over their sexuality. The company has not explained how it plans to prevent underage users from accessing the same features — a problem no tech company has solved so far.
Privacy experts have flagged the obvious risk. Users engaging in intimate exchanges with an AI agent that also has access to their email, social media and potentially financial information are exposing themselves to hacking and extortion. One privacy counsel described the situation as a “powder keg.”
The pattern is familiar – users formed similar emotional bonds with OpenAI’s GPT-4o last year, and revolted when the company tried to take the model down. But Muse raises the stakes. The difference between talking to a chatbot and handing a chatbot the keys to your inbox, your contacts and your credit card while also sexting it is not a difference of degree. It is a difference of kind.
Source: Futurism
