Amazon Mechanical Turk closes to new clients: the end of an era

What happened?

According to TechCrunch on July 5, 2026, Amazon has decided to stop accepting new requesters on Mechanical Turk (MTurk). The platform, launched in 2005, pioneered crowdsourcing of microtasks such as image labeling, transcription, and content moderation. Although existing workers (Turkers) can still perform tasks, the inability to onboard new clients drastically reduces workflow. The decision comes after years of declining data quality, criticism over low wages (often below the U.S. minimum wage), and competition from specialized platforms. TechCrunch also notes that Amazon has been redirecting resources toward AWS and generative AI services, leaving MTurk in the background.

Why is this important?

Mechanical Turk was not only an early experiment in the gig economy but also laid the foundation for the AI industry by providing human-labeled data to train models. Launched in 2005, MTurk was one of the first platforms to connect businesses with remote workers for tasks machines couldn't do. Its name comes from the 'Mechanical Turk,' a famous 18th-century automaton that was actually operated by a hidden human. For nearly two decades, MTurk was the backbone of data labeling for companies like Google, Microsoft, and AI startups. However, its generic, low-cost model has been overtaken by specialized platforms like Appen, Scale AI, and Prolific, which offer better quality control, ethical compliance, and fairer wages. Additionally, ethical criticism over low wages (often less than $2 per hour) and precarious working conditions pressured Amazon into this decision. A 2020 Stanford University study found that the median Turker wage was just $2.30 per hour, well below the federal minimum wage of $7.25. Amazon's decision reflects a broader trend: the automation of many microtasks (like transcription or basic labeling) via AI, and the growing demand for highly specialized workers for complex tasks, such as refining generative models.

Consequences

• For workers: The most dependent Turkers will lose a source of income. Although Amazon is not shutting down the platform, the reduction in projects will affect their activity. MTurk was estimated to have around 500,000 registered workers, though only about 100,000 were active. Many relied on MTurk as supplementary or primary income, especially in developing countries. The decision could push them to platforms like Prolific, which pays better (often between $6 and $10 per hour) but has less task volume. For the most vulnerable Turkers, the lack of alternatives could be critical, as many other platforms require specific skills or residency in certain countries.
• For clients: Companies still using MTurk will need to migrate to alternatives like Prolific, Clickworker, or generative AI platforms that integrate labeling. The migration won't be trivial: many companies had custom workflows on MTurk and will need to adapt to new APIs and quality control systems. Additionally, costs may rise, as alternative platforms often charge higher fees to guarantee quality and ethical compliance. For example, Scale AI charges between $0.05 and $0.50 per task, depending on complexity, while MTurk allowed fees as low as $0.01. However, data quality on MTurk was often poor, with error rates up to 20% on complex tasks, according to a 2018 study.
• For the market: Consolidation accelerates toward platforms offering better quality control and ethical compliance. The era of anonymous, low-cost microtasking is coming to an end. Companies like Appen and Scale AI have grown significantly in recent years, and MTurk's exit from the new-requester market could further boost their growth. Additionally, generative AI is creating new forms of human work, such as evaluating model responses (RLHF), which require more sophisticated and better-paid skills. This could lead to market segmentation: simple tasks automated by AI, and complex tasks performed by specialized workers on premium platforms.

What should readers know?

This move does not mean the death of crowdsourcing, but its transformation. New platforms prioritize transparency, fair wages, and more complex tasks. For AI developers, the lesson is clear: relying on a single source of human data is risky. Diversifying providers and considering synthetic or AI-generated data can mitigate risks. For workers, it's time to diversify skills toward platforms that value their contribution, such as those focused on content review, medical annotation, or language model training. Additionally, readers should watch for potential regulatory moves: the European Union is already discussing a directive on platform work that could require minimum wages and algorithmic transparency. Amazon's decision may be an early response to these pressures. In summary, MTurk's closure to new requesters marks the end of an era, but also opens opportunities for a fairer and more specialized human work ecosystem.