Amazon to Shut Down Mechanical Turk After 21 Years

Amazon has announced that it will permanently shut down AWS Mechanical Turk (MTurk) on September 30, 2026, bringing an end to a 21-year-old platform that helped businesses outsource small online tasks to human workers. Amazon says the decision followed an internal assessment of its programs, tools and services.

Mechanical Turk launched in 2005 and became an important part of the early crowdsourcing economy. Businesses could post “Human Intelligence Tasks,” including data labeling, transcription, surveys, content categorization and other jobs that were difficult for computers to perform reliably at the time. At its peak, the platform served more than 500,000 workers.

The closure comes as artificial intelligence has dramatically changed the types of work that originally made MTurk valuable. Tasks that once required large numbers of people can increasingly be handled by AI systems, while newer companies specializing in AI data labeling and human feedback have also entered the market. Amazon had already stopped accepting new MTurk customers in July, signaling that the service was moving toward the end of its operations.

There is an ironic element to the shutdown. When Mechanical Turk launched, Amazon founder Jeff Bezos described the concept as “artificial artificial intelligence” because humans were being used to perform tasks that computers could not yet handle effectively. More than two decades later, advances in AI have reduced the importance of many of those human-powered tasks. Research has also found that some MTurk workers were themselves using AI tools to complete assignments, further blurring the boundary between human and machine-generated work.

The end of Mechanical Turk therefore represents more than the closure of an old Amazon service. It illustrates how rapidly AI is changing digital labor. A platform originally created to provide human intelligence where computers fell short is disappearing at a time when AI systems are becoming capable of handling an increasingly broad range of those tasks. However, human workers remain important for specialized evaluation, difficult edge cases and high-quality data, meaning the future is more likely to involve a changing relationship between humans and AI rather than the complete disappearance of human input.