Do you ever feel like someone may be watching you in your own home? If you use an Amazon Cloud Cam home security camera, someone actually might be.
The online retail giant is being hit with allegations that it’s infringing on customer privacy with its security camera products. More specifically, a new report hints that the system may be subject to the same privacy concerns as Amazon’s voice-based product.
According to a new Bloomberg report, human Amazon workers may be reviewing Cloud Cam footage in order to train the company’s algorithms. In fact, “dozens” of employees in India and Romania may be given access to clips captured and stored by Amazon.
Those Amazon workers are tasked with reviewing and annotating certain clips captured by a Cloud Cam. They then let the system know whether those clips represent an actual security threat or if they’re simply a false alarm, five people familiar with the matter told Bloomberg.
Some of these Amazon workers can annotate about 150 video clips per day, each around 20 to 30 seconds long.
Amazon says the video footage is sourced from employee Cloud Cam testers, as well as consumers who have submitted clips to troubleshoot accuracy and quality issues.
The firm added that “only customers can view their clips,” unless they have been submitted for troubleshooting purposes.
Despite Amazon’s insistence that the clips are obtained from voluntary consumers, two of Bloomberg’s sources say that they’ve seen very private activity in the videos that Cloud Cam owners probably don’t want shared, including rare instances of people having intercourse.
For its part, the retail giant says that “inappropriate” clips are disposed of so that they don’t accidentally train the algorithm. She did not disclose whether those “inappropriate” clips were obtained from users who opted-in to troubleshooting
Amazon says it maintains tight security for its Cloud Cam, and workers in India annotate clips on a restricted floor where smartphones are banned. Despite that, one source told Bloomberg that some workers have shared footage with people not on the team.
If this all sounds familiar, it’s because it follows a similar controversy with voice-based digital assistants. Amazon, Google and even Apple were said to rely on human workers to review audio clips, some of which were accidental and contained sensitive material.