How did Roomba-recorded photos end up on Facebook?

The silver lining of is that now I can show this article to anyone who accuses me of being overly paranoid.

Key paragraph for our friends who don’t RTFA

> iRobot … confirmed that these images were captured by its Roombas in 2020. All of them came from “special development robots with hardware and software modifications that are not and never were present on iRobot consumer products for purchase,” the company said in a statement. They were given to “paid collectors and employees” who signed written agreements acknowledging that they were sending data streams, including video, back to the company for training purposes. According to iRobot, the devices were labeled with a bright green sticker that read “video recording in progress,” and it was up to those paid data collectors to “remove anything they deem sensitive from any space the robot operates in, including children.”

Seems like the real story is that training data was leaked, rather than the attention getting “they’re watching you” narrative the title suggests

The answer: special robots used during development for training ML image classification. Presumably leaked by human gig workers in Venezuela who were hired to perform image classification.

> All of them came from “special development robots with hardware and software modifications that are not and never were present on iRobot consumer products for purchase,” the company said in a statement. They were given to “paid collectors and employees” who signed written agreements acknowledging that they were sending data streams, including video, back to the company for training purposes. According to iRobot, the devices were labeled with a bright green sticker that read “video recording in progress,” and it was up to those paid data collectors to “remove anything they deem sensitive from any space the robot operates in, including children.”

I read the article and I don’t think your interpretation of the “suggested” narrative is there. The title is “ A Roomba recorded a woman on the toilet. How did screenshots end up on Facebook?”. That’s not really implying a “they’re watching you narrative” - who’s “they” and if it’s global syndicate why did they do something as innocuous as putting it on Facebook?

But yes, the real story is that training data (and “real” data) does leak all the time and that most companies don’t take insider risk as seriously as they should.

Scale AI contractors likely leaked these. Regardless of how they came to be posted onto Facebook, it still seems like iRobot's responsibility to keep this data under wraps.

We see similar stories all the time, whether it's about companies leaking data that was collected via consent, data collected without consent, or data collected without anyone knowing about it gets leaked.

Even Apple has been caught recording via Homepods without consent.