Amazon employees based in India and Romania have reportedly been watching videos taken from inside customer homes on the tech giant’s Cloud Cam home security camera, according to Bloomberg. The artificially intelligent device requires human workers behind the scenes to help train the AI algorithms to better detect threats.
The security camera allows people to keep an eye on their home 24 hours a day, seven days a week and is hooked up to Amazon’s Alexa voice assistant.
According to anonymous insiders at the company, workers are viewing hundreds of video clips per day, including some that are meant to be private. The employees commented that some of the video clips are intimate moments taking place inside people’s houses.
However, the company has stated that the video footage reviewed by Amazon employees is sent in voluntarily, usually to troubleshoot a problem, and that customers have complete control over their videos.
An Amazon spokesperson commented, “We take privacy seriously and put Cloud Cam customers in control of their video clips,” adding that unless the footage is submitted voluntarily, “only customers can view their clips.”
The terms and conditions of the Cloud Cam do not mention that human workers are training the AI algorithms to improve the motion detection software.
— Cord Cutter Deals (@CordCutterDeals) October 5, 2019
The anonymous employees also commented that when a clip includes private footage, it is automatically flagged and discarded. They added that despite this, some of the employees are still able to access clips meant to be kept private and in certain cases, share the footage with non-team members.
The Amazon spokesperson was unable to say why intimate activity would be showing up in voluntarily submitted clips.
Cloud Cam is available for just under $120 in the United States and films in 1080p HD. It detects and alerts users to activity going on in their house, giving them free access to the video footage for 24 hours. The technology includes extras that can be purchased separately, including extended access for up to one month and tailored alerts, such as for a crying baby.
The technology also hooks up to Alexa smart speakers, Amazon Fire TV, and various other products owned and produced by the company.
As AI algorithms are getting better at teaching themselves, tech companies employ humans to improve the algorithms and detect issues. However, using humans to train artificial intelligence inside consumer products has been a controversial subject among privacy advocates concerned about the exposure of personal information.
David Limp, who runs Amazon’s Alexa and hardware teams, regrets that the company has not been forthcoming about the fact that they use people’s data to develop AI.
“If I could go back in time, that would be the thing I would do better. I would have been more transparent about why and when we are using human annotation.”