Skip to content
NewsIncident

Amazon closes Mechanical Turk to new customers, sending the 21-year-old crowdwork platform into managed retirement

· by Pondero Newsdesk

The short version

AWS announced on July 30, 2026, it will stop accepting new Mechanical Turk customers, ending new signups for the crowdsourcing marketplace that trained a generation of AI models.

Amazon closes Mechanical Turk to new customers, sending the 21-year-old crowdwork platform into managed retirement

Amazon Web Services placed Mechanical Turk on its "Services in Maintenance" list on July 30, 2026, blocking new signups for the crowdsourcing marketplace that spent two decades supplying the labeled data that trained commercial AI models.

What

AWS confirmed the new-customer cutoff via an announcement on the Mechanical Turk website and in its SageMaker documentation. Existing customers can continue using the service. AWS stated it would keep investing in "security and availability improvements" but will introduce no new features, per TechCrunch's July 5 report.

The service launched in November 2005 as a marketplace where people completed small tasks that resisted full automation: flagging image content, identifying sentence sentiment, solving CAPTCHA challenges. Amazon billed it in 2018 as a data-annotation layer inside SageMaker, and it became a standard sourcing pipeline for machine learning training sets through the early 2020s.

The platform's final years were complicated by the very technology it helped build. A 2023 analysis cited by TechCrunch found that between 33% and 46% of Turk workers had begun using large language models to complete their assignments, raising questions about data quality and removing the human-in-the-loop rationale that the platform depended on. AWS did not give a specific reason for the maintenance decision beyond "careful consideration."

Why it matters

For AI tool operators, the shutdown marks a clean boundary in how labeled training data gets produced. The crowdwork model that Mechanical Turk embodied, paying workers fractions of a cent per task to produce annotations at scale, gave way to synthetic data generation and model-assisted labeling pipelines once frontier models crossed a threshold of reliability. SageMaker Ground Truth and automated labeling tools absorbed the use cases Mechanical Turk once owned.

The closure is also a signal about the data-labeling services market. Vendors that positioned themselves as higher-quality successors to Turk-style annotation, offering managed teams, quality audits, and domain expertise, may see demand from the Turk legacy customer base. Scale AI, Appen, and other labeling platforms had already drawn a large portion of enterprise annotation spend away from Mechanical Turk before the maintenance notice arrived.

For anyone still running annotation pipelines through Mechanical Turk, the service remains available for existing customers and AWS has not announced a final shutdown date. That decision, if it comes, is the meaningful operational deadline.

What to watch next

AWS has not disclosed when or whether it will terminate service for existing customers, only that July 30 closes the door to new ones. The practical question for Turk's remaining enterprise users is whether a formal end-of-life notice follows in the next six to twelve months and what migration path AWS recommends. SageMaker Ground Truth is the logical internal successor. Whether AWS bundles Turk's existing customer base into that service or simply lets the platform run down without a migration offer is the next development worth tracking.

Sources