Meta Files Patent for AI That Can Listen All Day and Track How You Feel
Meta has filed for a patent for an AI that listens to your voice throughout the day, assesses how it thinks you feel by the way you sound, and keeps a time-stamped log of every reading.
Each reading is pinned to the moment it happened: the time, your location, what you were doing, and how you were using your phone. Some versions in the file will listen all day; others only entered at appointed times.
None of these ships are in production today, and Meta has not announced one; Installations like this stake a claim long before anyone commits to building them.
The application, US 2026/0182881, was filed by Meta Platforms in December 2025 and published on July 2. It names one inventor, Lachlan Dunnand traces back to the interim filing from December 2024. Patent analysis site Patentlyze has flagged the filing first.
Its title combines two ideas, sentiment analysis and real-time fitness training. The claims show that the first one is the most important: out of 20, three independent ones include the emotional analysis themselves, while the exercise training is only visible in the dependent claims that build on them.
That’s what copyright defines
The device records your speech throughout the day. It could be smart glasses, a phone, a smart watch, headphones, or a smart home speaker, the patent says.
The device records it, and an AI trained to read mood starts working on both the words and the way you say them: your tone, your pace, sighs, laughs. It marks each part of the sound with a sentiment reading, matches the surrounding context, and over a set period of time, a day or a month, creates a summary of your patterns.
The system is not just a label for you to emphasize. It points back to the words behind each reading, what copyright calls a citation. In one example, angry reading comes with the harsh words you used.
One person puts one person through the day: idle language on a morning video call from home, a laugh with a friend at dinner, a 9:15 PM groan over a smart home speaker.
In that scenario, the speech patterns are “time-stamped and uploaded to the servers,” and the system provides the user with an example to read like this:
“He moans a lot before bed, and he’s happier when he’s with friends. He’s shown more gratitude this month.”
Applying goes beyond your word. It can roll with biometric signals and eye tracking, using pupil size, blink rate, and even eye moisture to mark stress or crying. It can also track how you use your devices, down to the posts you view or like, your screen time, and how quickly you switch between apps. All feed into one emotional profile.
The other half of the patent is an exercise trainer. The smart glasses watch your form in the mirror and tell you about the set, telling you that you’ve sunk deep into the square, then congratulating you over and over again.
The coach reads your feelings, too. If it senses that you are tired or depressed, it slows down. If it decides that you have the ability to save and you are lax, the patent says “it can order”. The patent claims that no human trainer can match its accuracy or keep it going all day.
We’ve been here before
Desire is not new. Amazon has added voice emotion reading to its wearable Halo in 2020. Its Tone feature listened to your tone and speed and told you how you experienced the day, calm, frustrated, and the like, and processed those samples on your phone and deleted them, never touching the cloud.
However, it has drawn scrutiny: in December 2020, Senator Amy Klobuchar pressed federal health regulators about Halo’s collection of verbatim body scan data, calling it extraordinarily strange.
Amazon shut down the entire line by 2023, though it never linked that to privacy. The gap with Meta installation is not really storage: the copyright stores the work on the device in some versions and the logs on the servers in others. It reaches. Tone reads your emotions with your voice alone; The Meta system also reads your eyes and your phone.
Regulators have their doubts about whether reading emotions this way works, and have begun to draw lines. As of February 2025, the EU AI Law prohibits AI that injects human emotions into workplaces and schools, without medical or safety reasons, with fines of up to 35 million euros or 7% of a company’s global turnover, whichever is higher.
Its editors flagged precisely a little science: the expression of emotions varies from person to person, culture to culture, and moment to moment. That restriction stops at consumer devices, however. A separate law coming in August 2026 will require systems that read emotions from biometric signals to disclose that they do so.
Whether the first voice coach is important is debatable; one that reads the pupils and the blink rate will fall more than twice that biometric line.
Hacker News has reached out to Meta for comment on whether the app features any branding programs and how such a program might handle user data, and will update this story with any response.
A fitness trainer is one job. Underneath it is a running log of everything the program has decided to experience, a key to where you’ve been and what you’ve been doing.
Amazon’s emotional learning listens to your voice alone, and it was pulled in 2023. Meta reaches your eyes and your phone too, and the only thing keeping it out of your life is that no one has built it.
