Meta Patents AI System That Tracks Your Mood Through Voice Recordings
Meta has been granted a patent for an AI system that continuously records a user’s voice, transcribes it, and feeds the audio through a machine learning model to detect their emotional state. The patent, published on July 2, describes a device that listens for “audible communications” including sighs, laughter, and vocal tone, and combines them with “contextual factors” like time of day, location, activity, and medication timing.
Key Takeaways:
- Continuous Monitoring: The system is designed for “continuous emotional monitoring on everyday devices.”
- Multimodal Inputs: It uses “multimodal sensor inputs on synchronized timelines,” combining audio with location, time, and other data.
- Fitness Tool: Meta positions it as a fitness tool for personal training, but its potential applications extend far beyond workouts.
- Privacy Concerns: Similar systems by Amazon (Halo Band) faced public backlash and were discontinued due to privacy worries. Meta’s planned use of voice recordings raises significant surveillance concerns.
Amazon tried something similar. In 2020, it launched the Halo Band, a fitness wearable with a built-in microphone that performed “tone of voice analysis.” After public backlash, Amazon removed the microphones from the next-generation model in 2021 and discontinued the product line entirely in 2023.
Meta’s smart glasses are already the subject of a privacy crisis over covert recording. Adding persistent ambient voice recording to its product line would extend surveillance beyond what users see to how they feel.
Subscribe to The 💜 of EU tech
Get the latest rumblings from the EU tech scene, insightful stories from our founder Boris, and some questionable AI art – all for free every week in your inbox. Sign up now!
A Meta spokesperson told 404 Media that “patents at Meta are often filed to disclose concepts that may or may not be implemented.” While this is standard corporate practice, the patent’s specific details on medication tracking and emotion-time correlations suggest something more than a mere filing.
If this patent becomes a product, it would involve training an emotional model on recordings of strangers’ sighs, laughter, and everyday conversations. This raises serious ethical questions about privacy and data handling.