Amazon’s Ring rolls out controversial, AI-powered facial-recognition feature tovideo doorbells

Amazon’s Ring doorbells can now identify your visitors through a new AI-powered facial-recognition feature called Familiar Faces. The company announced the controversial feature earlier this September, and it is now rolling out to Ring device owners in the United States.

Amazon says the feature allows you to identify people who regularly come to your door by creating a catalog of up to fifty faces. These could include family members, friends, neighbors, delivery drivers, or household staff. After you label someone in the Ring app, the device will recognize them as they approach. Then, instead of alerting you that a person is at your door, you will receive a personalized notification, such as “Mom at Front Door.”

Owners can use the feature to disable unwanted alerts, like notifications for their own comings and goings, and can set these alerts on a per-face basis. The feature is not enabled by default; users must turn it on in their app settings. Faces can be named directly from the Event History section or from the new Familiar Faces library. Once labeled, the face will be named in all notifications, in the app’s timeline, and in Event History. Labels can be edited at any time, and there are tools to merge duplicates or delete faces.

Amazon claims the face data is encrypted and never shared with others. It also says unnamed faces are automatically removed after thirty days.

Despite Amazon’s privacy assurances, the addition of the feature raises concerns. The company has a history of forging partnerships with law enforcement and once gave police and fire departments the ability to request data from the Ring Neighbors app by asking Amazon directly for doorbell footage. More recently, Amazon partnered with Flock, the maker of AI-powered surveillance cameras used by police, federal law enforcement, and ICE.

Ring’s own security efforts have fallen short in the past. Ring paid a five-point-eight million dollar fine in 2023 after the U.S. Federal Trade Commission found that Ring employees and contractors had broad and unrestricted access to customers’ videos for years. Its Neighbors app also exposed users’ home addresses and precise locations, and users’ Ring passwords have been found on the dark web for years.

Given Amazon’s willingness to work with law enforcement and digital surveillance providers, combined with its poor security track record, it is suggested that Ring owners be careful about identifying anyone using a proper name, or better yet, keep the feature disabled.

As a result of the privacy concerns, Amazon’s Ring has faced calls from U.S. Senator Ed Markey to abandon this feature and backlash from consumer protection organizations like the EFF. Privacy laws are preventing Amazon from launching the feature in Illinois, Texas, and Portland, Oregon.

In response to questions, Amazon said users’ biometric data will be processed in the cloud and claimed it does not use the data to train AI models. It also claimed it would not be able to identify all the locations where a person had been detected, from a technical standpoint, even if law enforcement requested this data. However, it is unclear why that would not be the case, given the similarity to the Search Party feature that looks across a neighborhood’s network of Ring cameras to find lost pets.

An EFF staff attorney commented that knocking on a door, or even just walking in front of it, should not require abandoning your privacy. With this feature going live, it is more important than ever that state privacy regulators step in to investigate, protect people’s privacy, and test the strength of their biometric privacy laws.