Meta has filed a patent application for an artificial intelligence system designed to continuously monitor a user’s emotional state by analyzing voice patterns, biometric signals, and device usage data. The patent application, identified as US 2026/0182881, was filed in December 2025 and published on July 2, 2026. The filing was first highlighted by patent analysis platform Patentlyze and names Lachlan Dunn as its sole inventor. Although Meta has not announced any plans to commercialize the technology, the application outlines a system capable of collecting and processing extensive personal data to determine how users are feeling throughout the day.
According to the patent, the proposed technology would function across multiple devices, including smartphones, smartwatches, headphones, smart glasses, and smart home speakers. The system would record speech, transcribe conversations, and use artificial intelligence models to analyze both the spoken words and vocal characteristics such as tone, speaking pace, laughter, and sighs. Every emotional assessment would be associated with contextual information, including the user’s location, the time of day, activities being performed, and how they were interacting with their devices. The patent also describes a process in which these emotional readings are stored and organized into summaries covering periods ranging from a single day to an entire month. In one example outlined in the filing, the system identifies behavioral patterns such as frequent sighing before bedtime or increased happiness when spending time with friends. The technology is also capable of attaching citations to emotional assessments by linking them to specific words or phrases used by the individual.
The proposed system extends beyond voice analysis by incorporating additional biometric and behavioral indicators. The patent describes the use of eye tracking data, including pupil size, blink rate, and eye moisture levels, to detect emotional conditions such as stress or crying. It could also analyze digital behavior, including the content users view or interact with, the amount of time spent on different applications, and the speed at which they switch between apps. These data points would contribute to the creation of a detailed emotional profile. Another feature described in the filing is a real time fitness coaching system. In this scenario, smart glasses would monitor exercise movements and provide spoken guidance and encouragement. The AI coach would also adjust its responses according to the user’s emotional condition, reducing the intensity of coaching when it detects fatigue or discouragement and becoming more demanding if it determines that a user is not exerting enough effort.
The filing has renewed discussions around privacy and the reliability of emotion recognition technologies. Similar concepts have appeared before, including Amazon’s Halo wearable, which introduced a voice based mood analysis feature in 2020 and was later discontinued in 2023. Regulatory authorities have increasingly scrutinized technologies that attempt to infer emotions through artificial intelligence. Since February 2025, EU AI Act has prohibited the use of emotion recognition systems in workplaces and educational institutions except for medical or safety related purposes. Additional regulations set to take effect in August 2026 will require systems that infer emotions from biometric data to clearly disclose their activities to users. While Meta’s patent does not confirm the development of a commercial product, the filing illustrates how technology companies continue to explore increasingly sophisticated methods of collecting and interpreting personal data through artificial intelligence systems.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
