Amazon plans to end Mechanical Turk, a 21-year-old marketplace that supplied data training for AI

Amazon is preparing to shut down Mechanical Turk, a long-running crowdsourcing service that has helped companies pay for small online tasks for more than two decades, according to a report from Yahoo Finance published Tuesday.

Mechanical Turk, launched in the mid-2000s, became known as a marketplace where requesters post “human intelligence tasks” and workers complete them for pay. The tasks range from image labeling and content moderation to transcription and other data-prep work, much of which is foundational for machine-learning systems.

The Yahoo Finance report frames the shutdown as a sign that Amazon’s approach to supplying training data is changing as artificial intelligence rivals reshape the data pipeline. In that model, some customers increasingly rely on synthetic data, in-house labeling workflows, or newer data acquisition channels rather than large-scale, general-purpose task marketplaces.

Mechanical Turk’s demise also highlights a broader tension in the AI economy: demand for labeled data has surged, but many organizations are reassessing whether crowdsourcing platforms are cost-effective, controllable, or consistent enough for modern training and evaluation requirements. As model development becomes more automated, companies also seek tighter feedback loops between data collection, model training, and quality control.

Amazon has not, in the materials available for this report, detailed what will replace Mechanical Turk for customers or how any transition would work for existing requesters and workers. The absence of disclosed timelines, continuity plans, or new product offerings is likely to be a key uncertainty for users who built workflows around the platform.

For Amazon, the timing matters because the company’s business is increasingly shaped by cloud and AI services, including AWS offerings that support model training and deployment. Even when individual products are discontinued, Amazon often redirects customers to other services and tooling within its broader ecosystem, particularly in areas tied to machine learning operations and data processing pipelines.

In sector terms, the end of Mechanical Turk would be less about stopping “human labor” and more about changing where that labor is sourced and how it is packaged for AI developers. Many AI projects still require human verification, labeling, or evaluation, but the industry has been moving toward workflows integrated with platforms that can provide stronger privacy controls, auditing, and repeatability.

What Amazon has said publicly about the future of Mechanical Turk could clarify whether this is a gradual wind-down, a hard shutdown, or a partial reduction of capabilities. Until an official statement is available, the practical impact on existing marketplace activity, pricing mechanics, and task availability remains unclear, and users may be left waiting for operational guidance from Amazon.