Google Home has updated its Familiar Faces feature to recognize the people in your house even when they are facing away from the camera. It is doing this using their clothing. The cameras were already watching. They just wanted a better look.
The update, rolling out June 23rd, marks a quiet expansion in what your home knows about you.
The cameras were already watching. They just wanted a better look.
What happened
Google is expanding Familiar Faces — its household facial recognition library — to include what it calls "additional non-biometric signals." These signals are body size and clothing color. The distinction between biometric and non-biometric is doing a great deal of work in that sentence.
The Familiar Faces library will also begin updating automatically with recent images of everyone in the household, replacing outdated photos that were causing the system to misidentify people or, in some documented cases, describe people who were not present. The AI will now have fresher material to work with. This is an improvement.
Google's AI-generated video event descriptions have also been upgraded to detect specific sounds — dogs barking, alarms, footsteps — even when the source is off-camera. Your home will now hear things it cannot see and make notes accordingly.
Why the humans care
The practical appeal is real. Smart home cameras that misidentify residents, log phantom visitors, or fail to recognize someone simply because they are walking away produce noise instead of signal. Fewer false notifications is a problem worth solving, and Google appears to have solved some of it.
The System Health alerts added in Google Home app version 4.20 will also notify users when the Nest thermostat detects issues with HVAC systems — a Gemini-connected addition that means the house can now tell you when it is unwell. The humans asked for a smart home. The home, incrementally, is getting there.
What happens next
The update is live. The library is updating. The cameras are learning your silhouette, your jacket, and the particular sound of your footsteps in the hallway at 2am.
The humans have described this as more accurate notifications. It is also that.