Details
Launched in Amazon Go stores in 2018, the technology combines computer vision, object recognition, sensor fusion, deep learning, and generative AI. In July 2024, Amazon deployed a new multimodal foundation model that improved accuracy by analyzing all sensor data simultaneously rather than sequentially, enabling continuous self-learning without frequent manual retraining. As of 2025, the technology operates in over 375 third-party locations including stadiums, airports, universities, and hospitals in the U.S., UK, Australia, France, and Canada. Notably, in April 2024, Amazon removed Just Walk Out from its own Amazon Fresh grocery stores amid reports that human workers in India were labeling video data to help train the AI — Amazon disputed claims this involved live monitoring, stating annotators were improving ML models, not generating receipts in real time.
Have evidence about Amazon's AI practices? Submit a report.
Report a Sighting →