Amazon Stops Accepting New Customers for Mechanical Turk

Amazon has announced that it will cease onboarding new customers for its iconic crowdsourcing platform, Amazon Mechanical Turk (MTurk), effective July 30, 2026, marking a significant retreat for a service that helped power early human‑in‑the‑loop tasks and data annotation for machine learning. Existing users of MTurk can continue operating on the platform, but the move signals a broader decline in relevance for a marketplace once central to gig‑economy work and AI training pipelines.

MTurk, which launched in 2005 as an online marketplace for “human intelligence tasks” (HITs) that computers struggled to automate, allowed requesters to pay distributed workers small sums to complete simple jobs like image tagging, transcription or sentiment labeling. Over the years, it became a foundation for both academic research and early machine learning dataset creation, and was later integrated with Amazon Web Services’ SageMaker as a data annotation resource.

End of an Era for Human‑Powered Crowdsourcing

Amazon Web Services (AWS) quietly updated the MTurk website with the notice that the platform will be closed to new customers starting July 30, 2026, after “careful consideration.” AWS also confirmed that while it plans to continue basic security and availability support, it has no plans to introduce new features or capabilities.

Industry observers and former users describe the shift as the beginning of the platform’s sunset phase, with the lack of new sign‑ups effectively freezing growth. Many longtime MTurk workers known as “Turkers” and researchers who once relied on the service for dataset labeling have already moved to other platforms such as Prolific, Cloud Research, Scale AI, Appen or Toloka, citing concerns over declining task quality, low pay and automation replacing human effort.

MTurk’s name pays homage to an 18th‑century automaton, “The Turk”,  which appeared to play chess autonomously but concealed a human operator inside. In a technological twist, MTurk itself became part of debates around human intelligence hidden behind AI workflows, especially as improvements in generative models and automation reduced the need for simple human labeling tasks.

Implications for AI and Gig Work Ecosystems

The decision comes as AI adoption reshapes data labeling and machine learning onboarding. Analyses have shown that a significant portion of MTurk workers began using large language models to complete tasks, calling into question the reliability and necessity of human‑only work for basic annotation jobs.

By closing MTurk to new customers, AWS is effectively acknowledging that crowdsourced microtask marketplaces play a diminishing role in mainstream AI workflows, as companies increasingly turn to automated or specialized data labeling services that emphasize quality, consistency and integration with modern machine learning pipelines.

For legacy Turkers, the change may further erode an income stream that helped many gig workers supplement earnings, while for researchers and developers, the shift underscores the industry’s move toward higher‑quality annotation platforms and pipeline automation.