Meta has patented a wearable AI system that would listen to a user and monitor their surroundings to infer emotional state, then use those conclusions to personalize fitness coaching. The filing, first flagged by Patentlyze, is a patent document, not a product announcement. Still, it is a revealing map of what Meta’s lawyers thought was worth claiming: continuous emotional monitoring on everyday devices.
According to the patent filing, Meta submitted the application in December 2025 and it was published July 2. The described apparatus would collect audio from a user and connect it with context such as time, location, activity and digital interactions. The system would transcribe speech and feed verbal and nonverbal signals into a machine-learning model that estimates emotional indicators.
The mechanism is straightforward in the way privacy problems often are. The device listens for speech, tone, laughter, sighs and other sounds. It also uses other sensor inputs and aligns them on a shared timeline. Meta’s filing says that pairing multiple streams of data would make emotional inference more precise and reliable.
The patent says the AI assistant could listen at defined times for communications including sighs, laughter and voice tone. It could use those inputs to quantify a user’s emotional state or produce other insights. One example in the filing says the system could associate a happier emotional state with a particular time of day or with the time medication is taken.
That medication example is doing a lot of work. The patent does not describe a mere step counter with cheerful coaching copy. It describes a device that may connect mood, routine, location, sounds and personal context into a profile of how the user feels over time.
What the system would collect
Meta’s filing says the model may need training data beyond spoken words. It describes data that can include attributes of thousands of objects, including books, personal messages and newspapers. It also says audible communications may include speech, voice data, sighs, laughter and other nonverbal sounds linked to expressions, emotions or ideas.
The stated use case is fitness. The filing argues that personal trainers lack the precision a machine could provide when correcting posture or movement. Meta’s proposed device would observe movement, recommend routines and provide corrective guidance from a single apparatus.
That framing does not shrink the surveillance surface. A wearable that records a user’s voice and surroundings would also capture other people nearby. The filing’s examples include emotional summaries and citations to specific audio moments that support the system’s interpretation, such as laughter on certain days, improved mood after life events or more positive emotion during morning routines.
Meta has already put cameras and microphones on faces through its Ray-Ban smart glasses. The patent points in the same direction: more sensors, more inference, more intimate data packaged as assistance.
The history here is not abstract. NPR reported that Facebook, Meta’s former corporate name, ran a 2012 “emotional contagion” study that changed the news feeds of about 700,000 users to test whether the platform could affect their moods. Users were not told they were part of that experiment, according to NPR.
Meta did not immediately respond to 404 Media’s request for comment. The company has not announced this patented system as a shipping product. A patent can sit unused for years. It can also mark territory for a company that has spent decades turning user behavior into ad-targeting machinery.
This story draws on original reporting from 404 Media.
