Amazon Mechanical Turk (MTurk) Shuts to New Customers: What Happens Next

AWS has announced that effective July 30, 2026, Amazon Mechanical Turk (MTurk) will stop accepting new customers and be placed into maintenance mode . The official AWS documentation states: "After careful consideration, we have made the decision to close new customer access to AWS Mechanical Turk, effective 7/30/26. Existing customers can continue to use the service as normal. AWS continues to invest in security and availability improvements for Mechanical Turk, but we do not plan to introduce new features" . AWS has added MTurk to its "Services in Maintenance" list, which signals likely eventual retirement .

Amazon has not publicly given a single definitive reason for the closure, but multiple contributing factors are clear from reporting and industry analysis .

Why MTurk Is Being Closed to New Customers

Growth of generative AI replacing human microtasks. Many of the small labeling and classification tasks that MTurk workers performed can now be done more cheaply and scalably by large language models and automated pipelines .

Data quality degradation. Independent research has shown declining data quality on MTurk, with bot infiltration and AI-generated responses contaminating results. A 2025 study found "critical shortcomings in current MTurk best practices" for survey recruiting . A 2023 comparison of five major online research platforms found MTurk ranked last across almost every quality measure tested . Competitor platforms like Prolific have published research showing MTurk's "data quality is much lower than that of competitors" .

Worker pool stagnation. Researchers have noted "slowing rates of population replenishment, and growing participant non-naivety" on MTurk, making it harder to get fresh, unbiased participants . The same workers tend to appear across studies, reducing the validity of results.

Ethical scrutiny. MTurk has faced years of controversy over low pay (tasks as little as $0.01), worker treatment, and mass account suspensions without explanation . Worker advocacy groups and journalists have documented how Turkers often earn below minimum wage .

No new feature investment. AWS itself signaled the platform's end by refusing to add new features beyond basic security patches .

How MTurk Evolved: From Microtask Marketplace to AI Training Tool

MTurk launched in 2005, originally built for Amazon's internal use — workers found duplicate product pages across Amazon's catalog . It grew into a general crowdsourcing marketplace for "Human Intelligence Tasks" (HITs) too difficult for computers, such as CAPTCHA solving, sentiment analysis, photo tagging, and audio transcription .

The pivot to AI training happened in the 2010s. The landmark ImageNet dataset — which kicked off the deep learning revolution in 2012 — was labeled by 49,000 MTurk workers from 167 countries . From that point, MTurk became the backbone of training data annotation for AI: identifying objects in images, labeling text sentiment, classifying content, and generating ground-truth datasets. At its peak, the platform had roughly 500,000 registered workers . AWS deeply integrated MTurk into its SageMaker AI ecosystem for ML ground-truth labeling .

By 2016, MTurk had become "a revolution in social and behavioral research," with over 1,200 studies published in a single year using the platform .

What This Means for Each Group

Existing MTurk requesters (companies and individuals already using the platform)

Can continue using the service "as normal" indefinitely, but will receive no new features and must operate within a frozen, maintenance-only environment . Growth of their workforce is constrained since no new requester accounts can be created.

Academic researchers

This is a major blow. By 2016, MTurk had already become a central tool for social science research . Researchers had already been migrating to alternatives due to declining data quality, bot problems, and participant non-naivety . The closure forces a permanent shift to competitors like Prolific, CloudResearch, or academic survey panels . Prolific, originally created at the University of Oxford, has been actively marketing to MTurk refugees with research showing better data quality and ethical treatment of participants .

AI companies that relied on MTurk for training data

Most large AI companies had already moved away from MTurk to specialized data annotation platforms (Scale AI, Labelbox, Sama, Appen) and synthetic data pipelines, which offer higher quality control and API integration . The closure formalizes a shift that was already underway. Companies still using MTurk for ground-truth human evaluation loops will need to port their workflows to these alternatives.

MTurk workers (Turkers)

The ~500,000 worker pool shrinks over time as existing requesters age out. Workers lose the ability to onboard with new requesters. Combined with Amazon's move to help workers "transition to the dedicated website," many Turkers will need to find work on other microtask platforms . Alternatives for workers include Clickworker, Appen, and Prolific .

The Bottom Line

MTurk isn't shutting down immediately, but its path to retirement is clear. For requesters, researchers, and AI companies still using the platform, the time to migrate workflows to alternatives is now. For the roughly 500,000 Turkers who powered the hidden human labor behind the AI revolution, the end of new customer access signals the beginning of the end for a platform that helped define the gig economy.