AI corporations have grown into data-hungry entities as their fashions require ever-larger datasets to coach on. To satisfy that want, many AI startups defy long-standing web conventions — like respecting robots.txt fileswhich sign to automated crawlers which components of an internet site are off-limits — and scrape data aggressively. This has pressured web sites to restrict access to their data and, in some instances, strike licensing deals with AI corporations. Health and social operating firm Strava is making a transfer on this path by limiting its web site and introducing charges for developer entry.
To cease scraping, the corporate is rising safety round its web site and can now solely enable authenticated customers to view sure knowledge. Earlier, customers had been in a position to see particulars like public profiles and health membership listings with out logging in. The corporate is placing all that knowledge behind authentication to guard it from unauthorized AI scraping.
On the API entrance, builders might beforehand begin constructing apps on Strava by a free, tiered entry program — making use of for fundamental entry first, then requesting extra as their app grew. Now the corporate is including a flat $11.99 per thirty days charge for all builders, although it famous the value might range by geography.
Strava mentioned its developer neighborhood has grown from 185,000 members final 12 months to 241,000 this 12 months, and the corporate plans to proceed supporting them. As a part of that, Strava additionally plans so as to add help for Mannequin Context Protocol (MCP), an rising commonplace that lets AI assistants and apps entry exterior knowledge in a structured manner, giving Strava extra management over precisely what will get shared and the way.
The corporate can also be planning to retire some API endpoints — discrete entry factors that permit exterior apps pull particular knowledge, like membership particulars — to guard consumer knowledge. Strava had already tightened API guidelines in 2024, banning its use for AI coaching and limiting third-party apps from displaying different customers’ knowledge. These adjustments drew backlash from builders who mentioned their apps can be severely affected.
Whereas some builders might settle for paying a subscription charge, sunsetting sure API endpoints might nonetheless impression dependent apps. Strava is giving builders a 90-day grace interval earlier than making these adjustments.
In an interview with TechCrunch, Michael Martin, Strava’s CEO, mentioned unchecked AI scraping might be the loss of life knell of the general public web.
“AI corporations are ruthlessly scraping public web sites, given their infinite want for coaching knowledge, which is degrading website efficiency throughout the board,” Martin mentioned. We’ve had a number of cases within the final a number of months the place efficiency has been diminished and, in some instances, impaired. Past scraping the general public websites, they’re additionally attempting to make use of our API to get entry to our knowledge, ignoring API phrases.”
He famous that Strava has refused overtures from main AI labs in search of knowledge licensing offers. He particularly singled out Perplexity, saying the AI search startup routed its scraping by aggregator providers to obscure its origin regardless of being turned away. That is in step with Perplexity having been accused of similar behavior elsewhere up to now.
Martin additionally flagged server overload brought on by poorly constructed vibe-coded apps, whose API calls are sometimes inefficiently structured and generate a disproportionate load on Strava’s methods. It’s a sample: when Meta banned third-party chatbots from WhatsApp final 12 months, it made an analogous argument about system overhead.
The timing in all probability isn’t coincidental. Strava confidentially filed for an IPO earlier this 12 months, and its transfer to guard its knowledge could also be supposed to sign knowledge self-discipline to potential buyers. The comparability to Reddit’s 2024 crackdown on API entry is one Martin was fast to handle. Not like Reddit, which priced API entry by the variety of calls (making it unaffordable for a lot of app builders), Strava is betting a flat charge retains the developer ecosystem intact.
“We wish the customers to really feel that they personal their knowledge and really feel snug with how we’re controlling and securing it. However we would like the builders to proceed to flourish and develop,” Martin mentioned.
While you buy by hyperlinks in our articles, we may earn a small commission. This doesn’t have an effect on our editorial independence.
