digital democracy|digital privacy
From tracking to training: AI adds privacy risks to workout apps
Personalized training is spiking in popularity, and so are AI alternatives that may be more affordable. But as technology promises to help you reach your goals, it also adds new risks to your personal information. This study uncovers the hidden cost of digital fitness — revealing that apps link the data they collect to your identity, track you, and now use it for AI training.
Key insights
- Google Trends reveals a clear pattern: the search term “fitness” spikes globally every January. Since 2022, the highest value was recorded in January 2026, reaching a score of 100. This score indicates peak search interest on a relative scale from 0 to 100, where 100 represents the highest interest during the chart's time period. On average, each January sees a 23% rise in search interest compared to the preceding December for each year in the analyzed period. April marks the start of the climb, building toward the next peak in summer. On average, the growth from April to the peak month in summer was approximately 13%. The January spike is likely driven by New Year's resolutions, whereas the increased interest in spring might be linked to people focusing on getting in shape for summer. However, researchers note that global physical inactivity levels haven't changed much in 20 years, with approximately 80% of adolescents and one in three adults worldwide not meeting the World Health Organization (WHO) physical activity guidelines.¹
- Technology, especially AI, is increasingly transforming the fitness industry and could shift how these challenges are addressed. By analyzing user data, AI has the potential to create highly personalized fitness experiences, tailoring workout plans to individual progress and goals. This demand is reflected in the increasing global interest in personal training, as indicated by Google Trends data, which shows notable growth in searches since 2025. To illustrate with numbers, the score in January 2025 was 37, while in winter 2026, it reached a peak of 100 during the analyzed period. This represents a 2.7-fold increase. Last year, the peak was in August, with a score of 75, and growth began in April. But this year, interest has been high right from the start, hinting it might stay strong all year long. While traditional personal training can be costly, AI may seem like a more accessible alternative.
- All the apps analyzed² incorporate AI features to improve user experience. However, with this advancement, such apps might also use personal data for AI development, which could lead to privacy concerns. For example, Strava uses gathered information from users to enhance the quality, reliability, and/or accuracy of their AI features by creating, developing, training, testing, improving, and maintaining AI and ML models run by Strava or its service providers.³ However, they state that, where possible, they use aggregated, de-identified information for AI features. In the case of Peloton, they use collected data to build, train, analyze, and improve the accuracy of their services, enhance products, and increase operational efficiency. While Peloton may use third-party AI service providers, they explicitly state that any personal data processed by these technologies is strictly for enhancing their services.⁴
- Among the top workout apps analyzed, Strava collects the most data linked to user identity, gathering 20 out of 35 data types listed in the Apple App Store. For example, these data types include location, purchase and search history, photos and videos, and other user content. Nike Training Club follows closely with 19 data types, while Peloton collects the least, with only 2 data types. Although many of these data types may be essential for app functionality, they can also be used for purposes such as advertising, analytics, product personalization, and more. For example, Ladder uses only 3 out of 10 data types linked to users for app functionality, but collects 7 data types for product personalization and employs 6 for analytics. Companies may also access and use additional sensitive biometric data when these apps connect to wearables or third-party services.
- Furthermore, 4 out of the 5 analyzed apps also use data for tracking, as stated by app developers in the information provided on the Apple App Store, with Apple Fitness+ being the exception. “Tracking” refers to linking user or device data collected from the app — such as a user ID, device ID, or profile — with user or device data collected from other apps, websites, or offline properties for targeted advertising purposes. Tracking also refers to sharing user or device data with data brokers.⁵
Methodology and sources
This study is divided into two main parts to explore fitness trends and the data collection practices of popular workout apps. The first part utilizes Google Trends to analyze search interest in “fitness” and “personal training” from January 1, 2022, onwards. This timeframe was selected due to enhancements in data collection since that date, allowing for a more accurate identification of global patterns and shifts in these topics over time.
The second part looks into how the five top workout apps for iPhone — Strava, Nike Training Club, Peloton, LADDER, and Fitness+ — handle data collection. These apps were selected from a CNET list² based on the largest number of monthly active users in 2025, as reported by Similarweb, with the exception of the preinstalled Fitness+, for which such data was not available. However, Fitness+ is likely used by most Apple device owners due to its default presence. We examined their data collection practices using information from the Apple App Store and reviewed their privacy policies for any details related to AI model training.
By combining these approaches, the study aims to provide a clear picture of current fitness interests and underscore the importance of data privacy in the digital fitness landscape.
For the complete research material behind this study, click here.
