Amazon shuts Mechanical Turk, ending 21-year gig platform

Amazon will close Mechanical Turk on September 30, 2026, ending a 21-year run that served more than 500,000 workers but has been overtaken by AI models and specialized data-labeling startups.

"Following an assessment, we've made the decision to close AWS Mechanical Turk, effective September 30, 2026," a notice posted to the platform's website said. Krista Pawloski, a data worker and organizer with Turkopticon, an advocacy group for data workers, said the platform had been "in decline" in recent years as Amazon invested fewer resources into improving it.

Mechanical Turk launched in 2005 to connect businesses with workers for "Human Intelligence Tasks" — data labeling, audio transcription, surveys — paying a few cents per task. Amazon founder Jeff Bezos called the service "artificial artificial intelligence," a nod to the 18th-century chess-playing automaton that concealed a human operator. The platform's decline accelerated as AI models improved and startups including Scale AI, Mercor, and Prolific entered the data-labeling market. A 2023 study by Swiss academic researchers found that up to 46 percent of Mechanical Turk workers were using AI models to complete their assigned tasks.

The closure affects insurance and travel companies that still rely on MTurk for data work, which are now racing to find alternative platforms. Pawloski, who started using the platform in 2008 and made it her primary job by 2012, said Mechanical Turk and similar platforms provide "meaningful income" and flexible hours comparable to ride-hail services or Amazon Flex package delivery. "There's some people that still pretty much still do it full time," she said. "They're concerned now."

From 500,000 Workers to Zero

Amazon initially built Mechanical Turk to help label large volumes of data on its webstore. The platform grew into a marketplace where businesses could post small digital jobs, with workers earning a few cents per task. At its peak, MTurk served more than 500,000 workers globally, making it one of the largest crowdsourced labor pools in existence.

The company later tried to reposition MTurk as a data annotation source for its SageMaker machine-learning service, seeking to capture demand from AI training pipelines. But as more specialized data-labeling services emerged, many turkers migrated to competing platforms, citing Amazon's perceived lack of investment in improving MTurk. The platform's user interface and payment systems remained largely unchanged for years, even as rivals offered better tools, faster payouts, and more consistent work.

The Paradox of AI Training Data

The irony of MTurk's closure is that AI both created and destroyed its market. The platform's data-labeling work helped train the very AI models that eventually made much of that work redundant. The 2023 study showing 46 percent of turkers using AI to complete tasks highlighted the paradox: workers were using the same technology that was displacing them.

The shutdown follows Amazon's decision last month to stop accepting new Mechanical Turk customers, a move many workers interpreted as a precursor to closure. The company said it regularly evaluates its programs and services and makes adjustments based on those assessments.

For the broader gig economy, MTurk's closure shows that AI-driven automation is now eliminating even the most basic digital labor categories. Data annotation startups like Scale AI, Mercor, and Prolific have absorbed much of the demand, but they too face pressure as AI models become more capable of self-supervision. The shift raises questions about the future of human-in-the-loop work, which remains essential for training AI systems on edge cases and domain-specific knowledge.

Amazon's decision to shutter MTurk is unlikely to move its stock — the platform was a small part of AWS — but it reflects the broader shift in the data-labeling market. The closure benefits competitors like Scale AI, Mercor, and Prolific, which have absorbed much of the demand for human-in-the-loop AI training data. For investors, the question is whether these specialized data firms can sustain growth as AI models become more capable of generating their own training data.

This article is for informational purposes only and does not constitute investment advice.