Ring cameras are getting an AI upgrade that can tell you what’s happening at your front door, as well as show you. Video Descriptions is a new feature that generates text descriptions of the motion activity on Ring doorbells and cameras.
Now, instead of an alert that says “Front door: person detected,” you’ll get something like “person with broom and mop is leaving.” Or instead of “Living room: motion detected,” you might get “a dog is tearing up paper towels on the rug.” You can see how this would be helpful; you probably don’t need to do anything about the first one, but the second one demands some action.
These new descriptive alerts will appear in the camera notification on your phone, so you can see at a glance if you need to bother clicking through and waiting for the video to load.
Video Descriptions is rolling out today, June 25th, in beta to Ring Home Premium subscribers in the US and Canada (English only), and will work on all currently available Ring doorbells and cameras, according to Ring.
In a blog post, Ring founder and recent new hire at the Amazon-owned company, Jamie Siminoff, said the feature is designed to deliver only the most relevant information. So, rather than a detailed description of the scene, the notification will focus on “describing the main subject that caused a motion alert and what action they are taking,” he said.
Ring is working on combining multiple motion alerts into one and developing custom anomaly alerts.
Video Descriptions joins Smart Video Search on Ring cameras, which launched late last year and lets you query your cameras about recent events via the app, such as “did the kids leave their bikes in the driveway?” Both AI-powered tools are available with the Ring Home Premium subscription ($19.99 a month), which also includes Ring’s 24/7 recording option.
Siminoff says Ring plans to use Video Descriptions to power more proactive home security features, including combining multiple motion alerts into one alert and, more ambitiously, to develop custom anomaly alerts.
This would “generate alerts only when something happens on your property that is an anomaly,” he said, explaining that Ring will be able to learn the routines of your home and only deliver notifications when something out of the ordinary happens. Meaning you might not get an alert for the person with a broom and mop leaving the house, but you will be notified that the dog is tearing up the living room.
Ring isn’t the only company using AI to improve camera notifications. Arlo recently launched AI-powered descriptions, which it calls Event Captions. Wyze also offers them, under the moniker Descriptive Alerts, and recently launched a No Big Deal filter that filters out all alerts other than those it deems high-priority. Google announced an AI descriptions feature powered by Gemini for its Google Nest cameras last year, but it’s only available in a public preview program and is still in a limited rollout. As with Ring, all the companies require you to sign up for a subscription for these features.
One big difference from Ring’s offering is that both Google and Arlo offer facial recognition, which Ring doesn’t. This should make descriptive alerts more useful. Getting a notification that “Johnny is opening a car door in the driveway” is more helpful than “A person is opening a car door in the driveway.”
The extra context text descriptions provide could be used by a gen-AI service like Alexa Plus to enable other actions in your home
Cameras see a lot of things you don’t need to know about, and notification fatigue is a real thing when it comes to alerts. Anything that can streamline and focus them is a good thing. AI-powered smart alerts for people, pets, packages, and vehicles were the first step, and now, more descriptive alerts with more information could make cameras more useful in the smart home. For example, the extra context text descriptions provide could be used by a gen-AI service like Alexa Plus to enable other actions in your home.
Of course, more information can lead to more privacy concerns. While detailed text descriptions of activity make it easier for people to keep an eye on their property, they also make it easier to keep track of people in a home, which could be abused by unscrupulous users.
Another concern is accuracy. Both of the descriptions themselves and of any custom filters Ring might offer in the future. If I don’t get the alert for the person with a mop leaving the house because the AI determined it wasn’t an anomaly, but actually it was a particularly fastidious burglar doing away with a fancy $700 Dyson mop, I’m going to be very annoyed.
