Strava recently informed its users and partners that new terms for its API restrict the data that third-party apps can show, refrain from replicating Strava’s look, and place a ban on using data “for any model training related to artificial intelligence, machine learning or similar applications.”
The policy is effective as of November 11th, even though Strava’s own post about the change is dated November 15th.
Effective November 11, the updated API agreement introduces three key changes that provide Strava users with greater control, security, and a consistent experience:
Stronger Privacy Standards: Third-party apps may now only display a user’s Strava activity data to that specific user. Users will continue to have access to their personal Strava data across apps connected to our platform, though there may be differences in how this data appears.
Data Use Limitations: Our terms now explicitly prohibit third parties from using any data obtained via Strava’s API in artificial intelligence models or similar applications.
Protecting the Strava Experience: Additional terms have been added to protect Strava’s unique look and feel and functionality, helping users easily distinguish between Strava and third-party platforms.
In a statement sent to The Verge, Strava VP of Communications and Social Impact Brian Bell said, “We anticipate that these changes will affect only a small fraction (less than .1 percent) of the applications on the Strava platform–the overwhelming majority of existing use cases are still allowed, including coaching platforms focused on providing feedback to users and tools that help users understand their data and performance.”
You can read the rest of Strava’s updated statement below.
Ray Maker, aka DCRainmaker, points out that Strava has widely served as a middleman to get data collected by wearables to other apps and services. According to Maker, “every single partner I’ve talked to over the last 24 hours has been completely broadsided by this change,” with their only notification arriving as an email on Friday giving a 30-day notice to make any adjustments.
Now, services like VeloViewer and Final Surge may need to find another way to provide the features Strava’s users explicitly ask them to provide, like automatically updated leaderboards, personalized workouts, and “year in review” recaps, even as Strava is beta testing its AI-generated “Athlete Intelligence” summaries that Maker says “tell me something vaguely positive but usually wrong.”
TrainerRoad is one app that promises to guide users through “Effective, Science-Backed Workouts,” based on access to data from Strava, and the team has already commented on the change. In a post on its forum, Nate Pearson writes, “This was crazy news to us, too,” but noted that Garmin users could use Garmin Connect instead and that the team had already been working on integrating other platforms like Zwift and Wahoo.
Brian Bell, Strava VP of Communications and Social Impact:
We wanted to provide some additional context around the changes to our API Agreement and the impact for our users and developers. We currently anticipate these changes will impact less than .1% of applications and proactively notified the majority of those affected last week.
Enhanced Privacy and User Control
Privacy and user control are at the forefront of our platform. As a result, we are committed to evolving our API practices as regulatory requirements and user expectations shift. This includes the decision to limit the ability of a user’s data to be displayed by third-party apps in ways the user may not expect. Specifically, we want to thoughtfully address situations where users connect to a third-party app and are unaware that their data is being surfaced not just for their own use and visibility, but also to other users (for example, in a public feed or heatmap). The latest API changes address this scenario and provide a more consistent framework for Strava user data.
Training AI Models
We believe in the potential of AI to transform the athlete experience–whether it’s delivering more personalized insights to help you reach your goals, generating route or training recommendations, or countless other possibilities. But innovation in this space must be handled responsibly and with a firm focus on user control. As part of our generative AI features, we are committed to implementing thoughtful solutions that prioritize user control and the ability to opt out.
Third-party developers may not take such a deliberate approach to training AI models and as a result, we believe the best decision for the platform and for users is to prohibit the use of data extracted from Strava users in this manner. Our previous terms already disallowed the use of Strava user data in model training and development but we’ve made this more explicit in light of the increasing activity in this space.
No Impact to Most Developers
We recognize that our platform thrives because of the creativity and dedication of third-party developers who build tools to complement and extend Strava’s capabilities. We are steadfast in our commitment to fostering this ecosystem. We anticipate that these changes will affect only a small fraction (less than .1%) of the applications on the Strava platform–the overwhelming majority of existing use cases are still allowed, including coaching platforms focused on providing feedback to users and tools that help users understand their data and performance.
There are plenty of posts on social media complaining about the sudden shift, but one place where dissent won’t be tolerated is Strava’s own forums. The company says, “…posts requesting or attempting to have Strava revert business decisions will not be permitted.”
Update, November 19th: Added statement from Strava.