What Happened
On February 23, MIT Technology Review published an investigation on the widespread but inadequately disclosed use of human teleoperation in humanoid robot development and consumer deployments. The report drew on academic research, company disclosures, and direct reporting to argue that the extent of human labor behind marketed "autonomous" robots is being systematically obscured.
The most specific example involved 1X, a Norwegian startup that has raised substantial funding for its Neo home robot, priced at $20,000 and announced for 2026 home delivery. The product includes an "Expert Mode" in which remote workers wearing Meta Quest 3 headsets control the robot through household tasks. The teleoperators see directly into customers' homes through the robot's cameras. 1X's founder acknowledged when asked that the company had not committed to any specific level of autonomous operation for the product.
Data collection practices were also documented: workers in Shanghai wearing VR exoskeletons and motion-tracking sensors spent weeks performing repetitive physical actions - opening and closing microwave doors hundreds of times per day - to generate training data for robot neural networks. Robotics companies have engaged workers at delivery companies to wear movement-tracking sensors during normal work shifts, with that data used to train manipulation policies.
The broader finding from academic research cited in the piece: when human workforces remain invisible in robot operations, the public and investors consistently overestimate machine autonomy. That overestimation drives investment decisions and product purchase decisions that would change if accurate capability levels were disclosed.
Why It Matters
Actual autonomy level determines total cost of ownership for any robot deployment. A robot that requires a human operator per shift is not autonomous labor replacement - it is remote labor with a physical avatar. The economics are completely different from genuine autonomy. Buyers and investors making decisions based on autonomy claims that don't reflect actual capability are exposed to significant financial risk when operational reality becomes apparent.
The privacy implications of 1X's Expert Mode are concrete: remote workers hired to help customers have direct visual access to their homes through robot cameras. Current marketing does not describe this arrangement clearly. Similar issues are likely to arise with other home robot products that use teleoperation as a quality or reliability enhancement layer.
For the robotics industry more broadly, the data collection practices described are legal and defensible as necessary research, but the opacity around them contributes to a trust deficit that will complicate future deployments, particularly in regulated industries or with privacy-sensitive consumers.
Our Take
The teleoperation disclosure gap is a genuine problem that will compound as deployment scales. Consumers purchasing a $20,000 home robot have a reasonable expectation of knowing whether a human employee will periodically be viewing their kitchen through a camera. That expectation is not currently being met by the marketing of some products in this category.
The data collection framing is more nuanced: capturing human motion data to train robots is standard practice in robotics research and is not inherently deceptive. The concern is specifically about lack of informed consent from workers whose motion is captured during normal employment and lack of disclosure to consumers about teleoperation involvement. A clear autonomy percentage labeling standard - stating what fraction of operating time involves human oversight or direct control - would address most of the legitimate transparency concerns without prohibiting the data practices themselves.