Amazon Mechanical Turk to Shut Down Sept. 30 After 21 Years

Mechanical Turk was built to supply the “human intelligence” computers could not provide. After 21 years, Amazon is shutting it down as that boundary becomes increasingly difficult to define.

Amazon Web Services said Mechanical Turk will permanently close Sept. 30, 2026, giving workers and businesses a little more than a month to move away from the platform. AWS did not provide a specific reason for the closure.

The shutdown follows Amazon’s July decision to stop accepting new Mechanical Turk customers and comes as AI systems take over some of the simple data tasks the platform once relied on humans to perform.

The human workforce behind AI

Launched in 2005, Mechanical Turk connected businesses and researchers with workers completing small “Human Intelligence Tasks,” or HITs. Jobs included data labeling, transcription, surveys, research, content moderation and product categorization, often for only a few cents per task.

The platform eventually became an important source of human-generated data for machine-learning projects. Amazon has said Mechanical Turk had more than 500,000 workers and offered access to a workforce operating around the clock.

Amazon founder Jeff Bezos, per CNBC, famously called the service “artificial artificial intelligence,” reflecting its original purpose: using people to handle tasks that computers could not reliably perform

But that relationship changed as AI systems became better at many of the jobs that once required human judgment.

A market that has moved on

Mechanical Turk now sits alongside a newer generation of data-labeling companies and labor platforms, including Scale AI, Mercor and Prolific, many of which focus more directly on AI training, evaluation and specialist human feedback.

Krista Pawloski, a data worker and organizer with Turkopticon, told CNBC that Mechanical Turk had been “in decline” as Amazon appeared to invest fewer resources in the platform while competing services gained ground.

The platform also faced an unusual problem as workers themselves increasingly turned to AI.

A 2023 study by Swiss academic researchers found that as many as 46% of Mechanical Turk workers used AI models to complete assigned tasks, according to CNBC. For companies seeking independent human judgment, that blurs the distinction between human-generated and machine-generated work that platforms such as Mechanical Turk were originally designed to provide.

More must-read AI coverage

What the closure means

For businesses, the immediate challenge is continuity. Some insurance and travel companies still rely on Mechanical Turk for data work, Pawloski told CNBC, and will need to find alternatives before the September deadline.

For workers, the impact is more personal. Pawloski said Mechanical Turk can provide flexible, remote income, including for people who rely on it as full-time work.

“There’s some people that still pretty much still do it full time,” Pawloski told CNBC. “They’re concerned now.”

The bigger shift

Mechanical Turk’s closure is less a sign that human data work has disappeared than evidence that the business around it has changed.

The platform was built around a simple trade: humans supplied the judgment computers lacked. Today, AI can handle more of those basic tasks, while the human work that remains valuable is increasingly being packaged by specialized companies for model training and evaluation.

That creates a dilemma for former MTurk customers. Moving to a specialist platform may provide more targeted AI-training services, but it also means leaving behind a long-running general-purpose marketplace.

Amazon has not named an official replacement for Mechanical Turk, leaving customers to migrate to other labor and data platforms while workers look for new sources of flexible online income.

The shutdown also closes a chapter in the history of AI itself. Mechanical Turk was created to supply the judgment computers could not yet provide; 21 years later, the line between human and machine work has become much harder to draw.

Other News: Bill Gates argues that some jobs should remain “Human Reserved” even if AI can perform them, pointing to areas such as childcare, education, and health care where human involvement may carry value beyond efficiency.