Young woman posed by a robot vacuum cleaner while sitting on the toilet. Images uploaded to the internet

A picture of a woman sitting on a toilet appeared on an internet forum in Venezuela in 2020. The picture was not taken by a person, but by a Roomba J7 robot vacuum cleaner made by iRobot.

The robot sent the images to Scale AI, a startup that hires people and companies around the world using smart devices to send it images, videos and audio files to train the artificial intelligence it is developing.

MIT Technology Review has uncovered several images and videos in which people can be easily identified. In addition to the woman sitting on the toilet, another image shows an 8- to 9-year-old boy whose face is clearly visible.

Roomba J7 robot vacuum cleaner.

The iRobot company was recently acquired by Amazon for $1.7 billion. The company confirmed that the images obtained by the source cited were captured by its devices in 2020. According to iRobot, they are from special versions of the robots, used for development. These units are not, according to the company, available to the general public.

Read:  Apple may not launch the next AirPods and HomePod until 2025

An advanced robot vacuum cleaner may be a danger to user privacy

The vacuums were given to paid partners and employees, who signed documents giving their consent for the vacuums to collect images, videos and audio content. This data is needed to make future Roomba vacuum cleaner models smarter, more accurate and more efficient.

Also read: What THeMIS does, the robot Ukraine got that Russia is paying a $16,000 reward for

The manufacturer claims that those who used the units, marked with yellow e-tickers that read “video recording in progress,” should have not allowed the robots into certain areas of the house.

Read:  Oppo Find N2 could be announced very soon with updated hardware

However, iRobot did not allow MIT Technology Review to see the agreements signed with partners and employees who used the robots. Moreover, the company did not provide contact information for the people who tested them.

What’s not known, so far, is how the images ended up on a public Internet forum. It could be a cyber attack that targeted Scale AI or iRobot, or they were put there by an employee of one of the companies who had access to them. And employees do have access to this data, as it has to be added to categories and annotated by a person before it is delivered to the machine learning algorithm.

The iRobot company has announced that it has opened an investigation into who made that data public, including who provided MIT Technology Review with 15 images.

The Best Online Bookmakers March 27 2024

BetMGM Casino

Bonus

$1,000