The Hidden AI Economy: When Health Apps Sell Your Data
On a Reddit forum dedicated to trading massive digital datasets, a user recently pitched an unusual goldmine: a database containing 150,000 labeled images of...

On a Reddit forum dedicated to trading massive digital datasets, a user recently pitched an unusual goldmine: a database containing 150,000 labeled images of human feces. The user noted that the dataset was "hard to obtain" and highly valuable for machine learning training.
This wasn't a bizarre prank, but a stark revelation about the hidden supply chain of artificial intelligence. The data came from an AI-powered gut health app called PoopCheck. Designed to give users a "daily gut health score," the app asks individuals to upload photos of their stool, which an AI then analyzes using the Bristol Stool Scale. Over the past few years, roughly 25,000 people used the app, unwittingly contributing to a massive, highly sensitive database that the app's founder was now trying to monetize.
The most alarming aspect of this incident isn't the nature of the images, but the glaring disconnect between the app's marketing and its actual legal framework. If you were to look at PoopCheck's website or its Apple App Store listing, you would feel completely secure. The marketing materials boldly claim "Privacy First," state that "your photos are encrypted," and explicitly note that "the developer does not collect any data from this app."
But the reality buried in the Terms and Conditions tells a completely different story. By creating an account and uploading an image, users legally grant the company a worldwide, irrevocable, and perpetual license to sell, distribute, and commercialize their data. The fine print explicitly states that the images can be used to train and commercialize AI technologies, and that these models can be sold to third parties. Furthermore, if a user decides they are uncomfortable and deletes their account, the agreement notes that any data already incorporated into AI models will not be removed.
This incident highlights a growing concern in the modern tech landscape: the insatiable hunger for AI training data is turning our most intimate moments into commercial commodities. As artificial intelligence models require increasingly vast amounts of niche data to improve, developers are finding creative—and sometimes deceptive—ways to harvest it.
For the everyday consumer, the lesson is clear but unsettling. In the era of AI, marketing promises about privacy are often a smokescreen. Until stronger regulations are enforced, the true cost of a free, AI-powered health analysis might just be your most private data, sold to the highest bidder.
Key Points
- The developer of an AI gut-health app attempted to sell a database of 150,000 user stool images for machine learning training.
- The app's public-facing marketing heavily promotes 'Privacy First' and claims no data is collected.
- The hidden Terms of Service legally allow the company to perpetually use, sell, and license user data for AI development.
- Data already fed into AI models cannot be deleted, even if the user closes their account.
Why It Matters
This case exposes the deceptive data-harvesting practices fueling the AI industry, proving that marketing promises of privacy often mask aggressive, legally binding data-selling agreements.
Sources:
更多专栏

The End of Car Buttons and CarPlay: How AI is Taking the Wheel
For the past decade, the ultimate fix for a clunky car dashboard was simple: plu...

The Agentic Divide: A Glimpse into AI's 2026 Landscape
What happens when artificial intelligence stops being a conversational novelty a...

The Physics of Siri: Why Apple's AI Dream Needs the Cloud
For years, the ultimate promise of smartphone artificial intelligence was strict...