A Roomba recorded a woman on the toilet. How did screenshots end up on Facebook?

3 hours ago

A Roomba robot vacuum captured intimate images, including a woman on the toilet, during development testing in 2020.
These images were sent to Scale AI for data annotation and later leaked by gig workers in Venezuela on social media.
iRobot confirmed the images came from modified development devices, not consumer products, with consent from paid collectors.
Data annotation involves human workers globally labeling images for AI training, creating privacy risks in the supply chain.
Robot vacuums with cameras collect invasive data; companies use real-home images to train algorithms for object recognition.
Privacy policies often allow broad data use for product improvement, but consumers may not understand the extent of sharing.
The incident highlights gaps in data protection, as facial images, including minors, were not consistently obscured.
The growing data annotation market relies on low-paid contract workers, making sensitive data vulnerable to leaks.
Legal frameworks in the U.S. lack comprehensive privacy laws, relying on company policies that may not fully protect consumers.
Robot vacuum companies aim for broader home robotics, increasing data collection needs and privacy concerns.

Hasty Briefsbeta