Shift’s free – Shift, an AI training startup owned by German lab MicroAGI, offers free home cleanings to people who let it collect egocentric video from a camera headset. The footage is de-identified and used to train AI for household robots—raising questions about privacy,
For anyone who agrees to a “free housecleaning” from Shift. the deal starts in plain sight: a cleaner shows up wearing a camera headset. But the moment that headset records the inside of your home. the transaction becomes something more intimate than most people expect—a first-person view of daily life that can later be licensed to train household robots.
Shift’s model is built around trading access to that data. Customers receive a free cleaning, while the people wearing the device—Shift calls them “operators”—are paid for participating. The company pitches the arrangement as transparent. describing it as a compensation bargain for data that. until now. has often been collected without clear rewards.
The company says the data is processed to blur identifying details. including names. faces. screens. ID cards. and other personal information. before it is incorporated into training datasets. Shift also says customers can withdraw consent and request deletion during a window before de-identification is completed. after which removal is limited. Still. even with those assurances. the questions that follow are difficult to ignore—because the recordings are made inside a private space. and because today’s “training data” can become tomorrow’s leverage.
Shift was launched by MicroAGI, a German data research lab founded in 2025. The company started by hiring contractors to record their own household tasks and then expanded the work: Poletaev says contractors expressed interest in recording more and contributing more. which included going into other people’s homes.
Recently, Shift began offering its cleaning services in New York City and across Europe. The company’s pitch to customers has a distinctly modern flavor—an easy, low-friction way to get work done now in exchange for becoming part of an AI training pipeline.
Anton Poletaev, cofounder of MicroAGI and co-CEO of Shift, frames the arrangement as an explicit trade. “We are very up front,” he says. “Yes, we are getting your data, but by doing so you’re finally getting rewarded for it, and you’re not being lied to.”
For the people who join the program, the promise is clear: participation comes with payment in the form of a free cleaning. But what’s harder to grasp is what the data can enable—especially once it leaves the home and enters the broader machinery of robotics training.
Shift wants diverse, high-quality footage from many homes, many layouts, and many variations of everyday tasks. Poletaev says diversity matters because a robot trained in one environment only may struggle in another. He points to exposure across lighting conditions. kitchen types. living rooms. and different taps that need repair—arguing that the training helps machines generalize across environments.
Technically, the recordings are captured using an egocentric video perspective: a camera headset records a first-person view of a cleaner’s hands. Shift says that hands-interaction perspective helps robotics systems understand how hands work with objects.
Shift is not alone in seeking this style of data. Startups such as Claru, Luel, Micro1, and Kled AI also offer contractor roles where people either film themselves doing tasks—like folding laundry or taking out the trash—or annotate datasets.
Shift, however, built its path from contractor filming to “operators” entering other homes. The company says its operators are vetted and trained. In New York, it also partners with existing local cleaning services, though it does not name those partners.
Scale is a key part of the pitch. Globally, Shift says it has collected hundreds of thousands of hours of data through dozens of thousands of operators. It says it has more than a thousand operators in the U.S., with the “vast majority” of recordings happening in New York.
Shift says de-identification is designed to reduce privacy risk. According to Shift, de-identification can happen within hours or up to a week after collection, depending on processing time. Harry Kilberg, the company’s U.S. general manager, says Shift is working to improve how it communicates that window.
The de-identified data is used for MicroAGI’s internal robotics research. Shift also says it may share data with “select robotics companies and frontier AI labs,” while insisting it is never shared publicly or used for advertising.
The company’s framing is that the bargain is fair: customers get a free cleaning, cleaners get extra pay for wearing the device. Shift says cleaners earn $20 an hour with “no fixed schedule.”
But the “win-win” argument meets a harder reality when the conversation shifts from consent to consequences.
Veena Dubal. a law professor at the University of California. Irvine. researches what she calls “precarious work. ” including platform workers. algorithmic management. and regulations around AI and work. She says the problem isn’t just whether people understand the exchange; it’s that they may not understand the downstream harms.
“ The average person doesn’t think about the downstream harms,” Dubal says. “And those downstream harms might not even be apparent to us for many, many years, and maybe they’ll be invisible.”
Shift says the footage isn’t used for advertising. Even so, Dubal’s point lands on uncertainty—because Shift isn’t the only company collecting this type of data, and not every company’s privacy policy is equally clear.
Claru’s privacy policy says it may collect personal information and share it with multiple third-party vendors. Kled’s notes say that if another entity purchases the content you submit. it may disclose biometric information. while adding that it does not “sell. lease. trade or otherwise profit from your biometric information.”.
That variation matters because the first-person home data Shift collects could, in a different future, be sold to brokers or retailers. Dubal raises a practical concern: personalized pricing for goods and services if companies know what’s in your home.
Then there is the issue that hits hardest because it’s so specific to the place where recording happens. Dubal asks what could happen if video captures something unlawful, like drugs. She points to a scenario where police might subpoena companies such as Shift as part of a criminal investigation.
“There’s just so much about us in our homes that we don’t even think about that when this becomes available, either to the public or private sector, kind of willy-nilly, it’s anyone’s guess how it could be used,” Dubal says.
She argues that the stakes change because the home is culturally, socially, and legally treated as a private space. Tech companies already collect information through phones, laptops, smart TVs, and other internet-connected devices. But filming inside the home captures data that those devices may not: how people move. how they live. and what they do when they are not on a phone or computer.
“It’s a radical shift,” Dubal says. “There is something dramatic about the idea that even this space is open to the market.”
Dubal also doubts the strength of anonymity promises. She says de-identification remains contested and that debate often centers on what counts as “personal data.” She points to the European Union’s General Data Protection Regulation. which defines “personal data” as “any piece of information that relates to an identifiable person.”.
“Firms will always say, ‘Oh, well, we’re not using your name,’” Dubal says. “But the reality is that they have so much data that they can figure out who you are without using your name.”
The concerns don’t stop at customers’ homes. Shift is also collecting data about the cleaners it hires—data that could eventually help replace workers.
Dubal warns that the information could be used not just to train robots, but to create software that controls workers in new ways—setting efficiency standards, consolidating jobs, and pushing people to work harder and faster for less, leaving housecleaning “like an Amazon warehouse.”
Ai-jen Poo, president of the National Domestic Workers Alliance, echoed that concern in a statement to Fast Company: House cleaners’ labor should be “respected and protected, not treated as background inputs for someone else’s technology product.”
Supporters of AI often argue it will democratize access to technology and services. Yet the tech industry’s own track record. Dubal says. shows that those promises can arrive with major disruptions—like Uber. which displaced taxi drivers and used data to train self-driving systems. before prices rose beyond its early VC-subsidized days.
That leads to the question many workers—and customers—may be asking underneath the marketing: will household robots actually be affordable and accessible to all?
Dubal says the goal itself is misplaced. “It’s not that we all need servants,” she says. “It’s that we all need jobs that pay well.”
Poletaev sees a different end point. He says the data Shift collects is driven by a desire for a world where everyday goods and services are abundant and accessible. In the meantime, he insists Shift will “make sure people are compensated throughout this transition.”
For Shift. the bargain is straightforward: people can get paid for their data now while helping build a future that may ultimately require less of their labor. For Dubal and the workers’ advocates she points to. the bargain’s cost is less visible—because it may arrive slowly. through privacy erosion. new forms of control. and the shifting value of work that once depended on human hands.
Shift MicroAGI AI training data egocentric video robotics household robots privacy de-identification labor gig work domestic workers EU GDPR