Amazon confirmed it will stop accepting new customer registrations for Mechanical Turk, its 18-year-old crowdsourcing platform that connects businesses with gig workers to complete micro-tasks. The company did not announce plans to shut down the service entirely, but halting new signups signals the platform's declining strategic importance to Amazon's business.
Mechanical Turk launched in 2005 as an experimental marketplace where requesters could post small jobs like data labeling, content moderation, and surveys for workers to complete for modest pay, typically ranging from pennies to a few dollars per task. The platform became foundational to AI training datasets and academic research. Universities and tech companies relied on Turk workers to annotate images, transcribe audio, and validate machine learning models at scale.
The decision reflects Amazon's shift away from human-powered crowdsourcing toward automated solutions. As machine learning capabilities improved, demand for manual labeling declined. Amazon's own investments in computer vision and AI reduced its internal reliance on the platform. AWS customers exploring alternative annotation services like Scale AI and Labelbox have pulled budget from Mechanical Turk.
Worker advocates had criticized the platform for years. Critics pointed to low wage standards, where workers earned below minimum wage when accounting for time spent. Amazon faced pressure from labor organizers and academic researchers who questioned the ethics of Mechanical Turk's compensation structure. The company took minimal action on these complaints.
Existing customers and workers can continue using the platform. Amazon did not set a shutdown date, leaving the door open for potential revival or gradual deprecation. For workers who depended on Turk as supplemental income, the announcement removes a key option in the gig economy landscape.
The closure signals Amazon's broader retreat from experimental consumer services that no longer align with core priorities. Mechanical Turk was always positioned as a test project rather than a pillar