Amazon Mechanical Turk closes to new customers July 30, 2026

What Amazon announced

On July 5, 2026, TechCrunch reported that Amazon Web Services will stop accepting new customers for Mechanical Turk on July 30, 2026. The banner is already live on mturk.com, and existing users are unaffected for now (TechCrunch, July 5, 2026). The official notice sits at the top of mturk.com. The Register reached the same conclusion two days earlier (The Register, July 3, 2026).

MTurk launched in 2005 as the original human-in-the-loop labor marketplace, coined "artificial artificial intelligence" by Jeff Bezos. It powered a generation of academic surveys, dataset labeling, and content moderation. Generative AI absorbed most of that workload over the last three years, and AWS now steers data-labeling customers to SageMaker Ground Truth, which layers human review on top of model-generated labels rather than the other way around.

For Amazon sellers this is not an operational hit today. Very few brand teams use MTurk directly. It matters because it is the clearest signal yet that Amazon considers the "cheap crowdsourced human" primitive obsolete, and that AI-plus-review pipelines are the default path forward for every data ops workflow that used to route through it, from product-attribute enrichment to review triage.

Key facts

Timeline

Why this matters for Amazon brands

• •AWS data-labeling roadmap is now Ground Truth first. If any part of a brand's catalog enrichment, review moderation, or PDP QA workflow still relies on MTurk, plan the migration to SageMaker Ground Truth Plus or a specialist vendor before the next AWS re:Invent cycle.
• •Human-in-the-loop shifts to review, not do. The economically viable pattern is now LLM does the first pass, a smaller human pool reviews edge cases. Any operational SOP that budgets for full manual coverage overestimates cost and underestimates speed.
• •Brand-safety and content-moderation vendors get a fresh look. The workers who ran review-triage tasks on MTurk will disperse across specialist platforms. Expect pricing to move in both directions over the next two quarters.
• •The bigger signal: Amazon is comfortable retiring a 20-year-old service the moment its core use case is absorbed by AI. Every other manual-labor SaaS layer sitting between a seller and Amazon is now on the same clock.

Related coverage

• Amazon and OpenAI strategic partnership - the other side of AWS's AI pivot.
• Amazon AgentCore opens to external services - AWS's agentic infrastructure play.
• Amazon's AI listing tools - AI-plus-review in the seller-facing surface.

How Nova helps

Nova is the operating system for Amazon brands. As Amazon retires the manual layer and pushes AI-plus-review everywhere, the operators who stay ahead are the ones with clean, reconciled data feeding their own AI workflows.

• Seller Cockpit - 200+ Amazon metrics reconciled across 21 marketplaces, ready to feed any downstream AI pipeline.
• Nova MCP for Claude - gives Claude live, reconciled Amazon context so brand teams stop copying spreadsheets into prompts.
• Amazon AI Agents - purpose-built workflows for the tasks brand teams used to outsource to human labor pools.