As Amazon, Google and Apple add familiar-face features, cameras can do more than record visitors: they can identify and log people who have never agreed to be scanned.
A delivery worker on a suburban round does not usually think of herself as a data subject. She thinks in stops, parcels and timings. Yet the newest generation of smart doorbells can do more than record her arrival and departure. They can match her face against a stored profile, keep that profile for later use and allow the household to add a name. That shift, now spreading through products from Amazon, Google and Apple, is moving smart home cameras from simple recording into a more contested form of biometric identification.
The issue has already reached the courts. According to recent reporting by CNET, three proposed class actions filed in the United States accuse Amazon and Google of creating facial templates from people who never consented to being scanned. The lawsuits focus on Ring and Nest devices with familiar-face features, alleging that delivery workers, neighbours and other passers-by are captured, analysed and stored as identifiable profiles. The companies have disabled the feature in jurisdictions with stricter biometric rules, including Illinois, while presenting the processing as secure and local or otherwise protected by their systems. Apple, meanwhile, entered the same market on 13 October 2026 with cameras and a video doorbell that it says can recognise familiar faces.
At the centre of the claims is a basic asymmetry. The people bringing the cases did not buy the devices, yet their faces are the ones being processed. One complaint filed in the Western District of Washington says Ring’s Familiar Faces feature scans guests and bystanders and keeps the resulting data in Amazon’s cloud. The Google cases, later consolidated in the Northern District of California, make a similar point about Nest: the technology is activated by the owner, but the person being scanned has no practical way to consent, object or even know which camera captured them. The pleadings are still untested, but they expose the practical limits of asking a homeowner to secure permission from strangers who merely pass by.
The companies’ own guidance shows they understand the consent problem, even if they address it imperfectly. Google tells owners to comply with the law and says they may need consent to identify people visiting the home. Ring’s help page makes a similar warning and says the feature is unavailable in several places, including Illinois, Texas, Portland in Oregon and Quebec. But those instructions place the burden on the household, not on the company or the person whose face is being processed. A homeowner can ask a relative or a regular visitor for permission. A courier on a route, or a stranger crossing the pavement, cannot realistically be approached in advance.
The claim that the data stay “on device” also does not settle the wider question. Ring says its familiar-face profiles are encrypted and stored in the cloud, not on the camera itself, and that unnamed profiles are removed after 30 days without recognition. Google’s support materials are more mixed: some older Nest setups store familiar-face data in the cloud, while newer ones use device memory with cloud support for clean-up. Apple’s marketing is different again. It says analysis happens locally on the customer’s Home hub or in Private Cloud Compute, a system Apple says does not retain user data once the response is returned. Those technical differences matter for security and company access, but they do not alter the central fact that the owner can create a named record of a recurring visitor.
Apple’s arrival makes the broader trend harder to dismiss. The company is selling the feature as part of a privacy-conscious smart home range, with cameras and a doorbell that can recognise people, pets, vehicles and packages. Its own user guidance allows owners to add a name or relationship to unidentified people captured by the camera, even if the label is something as plain as “The Postman”. That may limit what Apple itself can see, but it still leaves the household free to build a local directory of people who come and go. In the United Kingdom, that matters further because domestic cameras that look beyond a property boundary are already subject to data protection rules.
The legal backdrop in Britain and Europe is more demanding than many consumers may realise. The Court of Justice of the European Union held in Ryneš that a household camera capturing part of a public footpath is not covered by the domestic-use exemption. The Information Commissioner’s Office gives similar advice, telling home owners to aim cameras at their own property where possible and warning that if a device records public areas or someone else’s land, data protection law applies. In practice, that can mean signage, limited retention, responses to access requests and a lawful basis for recording. The English courts have already shown, in Fairhurst v Woodard, that a Ring doorbell can be acceptable for incidental capture but problematic when it goes further than is necessary.
The most sensitive part of the debate is not just recording, but recognition. A camera that sees a stranger at the door is one thing. A camera that remembers that stranger, attaches a name and treats every return visit as part of a pattern is something else. Research by Data & Society found that doorbell cameras are already used by customers to monitor, instruct and sometimes discipline delivery workers, turning a doorstep into a workplace as well as private property. Recognition intensifies that relationship. It turns each stop into a searchable event and makes the ordinary movement of a courier, cleaner or neighbour more legible to the household that owns the device.
That is why the question is now larger than any one lawsuit. The industry is converging on the same idea from different directions: a smart doorbell should not merely watch the street, but identify it. Amazon, Google and Apple have each built systems that can convert an unknown face into a named entry. The technical implementation varies, but the social effect is similar. The presumption that a person can pass a front door without being filed away is being replaced, product by product, with a design that treats recognition as a normal feature of domestic security.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
