Technology
Amazon delivery glasses will capture images thousands of times per shift—with little transparency
Smart glasses being tested by Amazon drivers photograph surroundings constantly to feed AI systems, raising privacy concerns that the company hasn't fully addressed.
The Verge
Amazon is equipping delivery drivers with smart glasses that will photograph their surroundings throughout their workday—potentially capturing thousands of images per shift—and send that data to the company's AI systems for processing.
According to reporting by Bloomberg, a single driver wearing the eyewear could collect "several thousand captures in a typical shift." Amazon's Viraj Chatterjee told the outlet the company plans to feed this visual information into its AI Wellspring platform, ostensibly to streamline the delivery experience. When pressed about whether customers might opt out of appearing in these photos, Chatterjee's response was telling: "We haven't thought about that."
Privacy protections remain incomplete#
Amazon says it will apply automatic blurring to detect people and vehicle license plates before images go to human reviewers. The company also confirmed that law enforcement could access the photos with a warrant. However, significant gaps remain in the privacy framework: Amazon has not disclosed how long it will retain the images, and customers have no mechanism to view or request deletion of photographs taken of them or their properties.
Growing surveillance concerns among drivers#
The smart glasses represent an escalation in Amazon's use of monitoring technology for its delivery workforce. Drivers already contend with AI-equipped cameras inside delivery vehicles—systems that have drawn complaints for incorrectly flagging unbuckled seatbelts and even capturing personal moments like coffee breaks. Prior incidents have also seen vehicle footage leaked publicly, adding to driver skepticism about data handling.
Expansion plans ahead#
Amazon is scaling up deployment, aiming to put 20,000 additional pairs of the glasses in service by the end of 2027. Drivers in ongoing pilot programs have already completed more than 275,000 deliveries while wearing the devices, generating a substantial dataset that feeds the company's machine learning infrastructure.
