Amazon will stop accepting new customers for Mechanical Turk, its crowdsourcing platform, on July 30, 2026. The company announced the decision on the platform's homepage and in AWS documentation, citing "careful consideration." Existing customers can keep using the service, but Amazon has signaled it will not add new features — only maintain security and availability.

Mechanical Turk launched in 2005 and became the dominant marketplace for human intelligence tasks, or HITs: image labeling, transcription, sentiment analysis, and other work that resists full automation. For academic researchers and companies building machine learning datasets, it was the go-to infrastructure for decades.

Not a Shutdown — A Managed Exit

This distinction matters. AWS is not killing the platform outright — it is stopping new signups while letting current users continue. Existing pipelines will not break on July 31. However, the combination of a hard cutoff for new customers and no future feature development puts Mechanical Turk firmly in maintenance mode. In practical terms, teams treating this as temporary or reversible would be taking on risk.

The language around "security and availability improvements" follows a familiar pattern in the tech industry: it signals a service is winding down without a specific end date. Any team building long-term workflows on Mechanical Turk should plan for eventual discontinuation.

Why Now? The AI Shift

The timing aligns with a major change in how machine learning training works. The workforce Mechanical Turk assembled — and the tasks it performed — largely overlaps with what generative AI and modern machine learning tools now do internally. Large AI companies generate synthetic training data at scale; newer models handle classification and transcription tasks that once required thousands of human annotators. The crowdsourced data-labeling market has not vanished, but it has split. High-end enterprises now use specialist vendors with vetted, managed teams, while synthetic data approaches handle commodity work that used to go to general crowdsourcing platforms.

Amazon itself has shifted its data-labeling infrastructure toward Amazon SageMaker Ground Truth, which supports multiple workforce channels, including Mechanical Turk. With the public crowdsourcing option closing to new users, Mechanical Turk's role becomes a legacy pathway rather than a growth platform.

For academic researchers — especially in fields like computational social science and behavioral economics — the impact is real. Teams have relied on Mechanical Turk's open enrollment to recruit diverse research participants. Platforms like Prolific and CloudResearch, built specifically for this segment, are the most likely to absorb displaced users.

Corporate Restructuring Context

The announcement coincides with Amazon's broader pullback. Reuters reported that Amazon confirmed 16,000 corporate job cuts in January 2026, completing a reduction of roughly 30,000 corporate positions since October 2025. Mechanical Turk's managed decline fits a pattern of the company consolidating its portfolio and directing engineering effort toward higher-growth products. The company has not explicitly linked the two, but the timing is consistent.

Mechanical Turk was named for an 18th-century chess automaton that concealed a human inside — a joke at automation's expense. The platform existed to coordinate the human labor that machines could not yet do. Now the automation Mechanical Turk helped train is capable enough to replace much of that work. The quiet retirement of the service, rather than a dramatic replacement, says something clear about where the industry stands.

Teams currently on Mechanical Turk have a practical next step: assess how dependent their data pipelines are before July 30 and map out alternatives — whether Amazon SageMaker Ground Truth with a managed workforce, a specialist labeling vendor, or synthetic data generation where the task allows it.