Business Insider has reported that Uber recently notified a select group of its gig-based contributors—specifically those engaged in artificial intelligence (AI) training assignments—that their services would be discontinued earlier than anticipated. These workers, who had originally been assured several remaining weeks of employment, were informed two full months before their scheduled end date that their participation was no longer required. This unexpected termination came as part of Uber’s broader internal restructuring of a venture known as *Project Sandbox*, the codename for the company’s initiative in providing AI development support on behalf of Google. Project Sandbox constituted one of Uber’s initial attempts to extend its technological expertise beyond its signature transportation and delivery ecosystem, aiming instead to deliver advanced AI tooling and training solutions for external corporate clients under the auspices of its emerging AI Solutions division.
According to two contractors familiar with the situation, roughly a dozen independent workers had been assigned to Project Sandbox at its peak, though the precise number of individuals ultimately released remains ambiguous. Early Monday morning, Uber sent an email to affected participants, explaining that the project’s chief client had recently altered its internal objectives and strategic priorities. This shift, Uber said, directly undermined the necessity for certain ongoing AI training operations. The message went on to assure those impacted that Uber remained committed to maintaining professional ties with them, promising to notify them in the future should new projects arise that align with their expertise and professional backgrounds.
Three contributors who discussed their experiences with Business Insider described themselves as highly qualified specialists, having attained advanced academic credentials such as doctoral degrees or other relevant scientific or technical work experience. These workers had joined Project Sandbox during the early autumn season with the understanding, based on Uber’s representations, that they would remain engaged for a minimum period of three months. Despite those expectations, their contracts were truncated unexpectedly, illustrating the fragile and often unpredictable nature of contingent labor arrangements in the technology sector. When approached for clarification, an Uber representative declined to issue any additional commentary regarding the decision or its broader implications.
Over the years, Uber has focused heavily on consolidating its global dominance in core business areas—namely, its ride-hailing and food delivery operations. These divisions have reported healthy double-digit gains in gross bookings during the company’s most recent fiscal quarter, and Uber’s stock value has jumped approximately 39% in 2025 to date. Yet, beneath the surface of these established enterprises, the company’s AI-oriented division remains in its nascent stage. Last month, Uber disclosed plans to initiate a limited U.S.-based trial of its new *Digital Tasks* program, a novel initiative enabling gig workers—including active drivers and couriers—to assist in data-centric assignments that contribute to AI model training.
Chief Executive Officer Dara Khosrowshahi explained during Uber’s latest earnings call that certain components of this Digital Tasks program demand advanced educational qualifications, including PhD-level expertise. He positioned these roles as a reflection of Uber’s ambition to evolve beyond its traditional identity as a transportation facilitator to become a versatile, technology-driven platform for diversified digital labor. This expanded vision situates Uber as an intermediary not only between passengers and drivers but also between organizations requiring human input for machine learning refinement and an increasingly flexible global workforce eager to provide it.
One Project Sandbox participant recounted to Business Insider that initial recruitment began when a staffing agency distributed an unsolicited email advertising the opportunity earlier this fall. After successfully completing an evaluation process that confirmed their technical suitability, the contractor was cleared for onboarding, with official documentation guaranteeing at least a three-month engagement. Following integration into Uber’s systems, each worker was paired with a designated liaison from Google—the client enterprise receiving the AI training output. Representatives from Google, however, declined to respond to journalists’ inquiries regarding their role or perspective on the project’s termination.
In practice, the tasks assigned to Sandbox workers were diverse and technically detailed. Responsibilities spanned from labeling and annotating images or video sequences to critically assessing whether AI-generated responses to specific prompts were both accurate and contextually sufficient. Compensation among contractors varied significantly. Internal materials reviewed by Business Insider indicated a pay structure beginning at $55 per hour and extending to as much as $110 per hour under optimal working conditions—equating to roughly $19,000 in monthly earnings for those who maintained a full 40-hour weekly schedule. Nevertheless, several contributors stated that, at times, Uber limited their weekly hours below the threshold required to earn top-tier rates, making such potential earnings largely theoretical.
According to the latest correspondence, Uber informed contractors that their alloted assignments would officially close on Tuesday, and the next procedural step would involve returning the laptops provided for project-specific use. As of the report’s publication, some workers still awaited their initial paychecks, with company communications indicating that payments could take as long as seven weeks from each worker’s onboarding date to process fully. One veteran of similar technology gigs remarked that, in his experience, most other companies he had previously contracted with honored the full duration of promised agreements, underscoring his disappointment at the abrupt conclusion of Project Sandbox.
This episode unfolds within a much larger economic context: data labeling and other forms of human-in-the-loop AI training have rapidly matured into an expansive, globalized subcontracting market. Tens—or even hundreds—of thousands of individuals worldwide now derive income by tagging data points, verifying algorithmic outputs, and performing the intricate behind-the-scenes labor required to make artificial intelligence systems more accurate and contextually intelligent. Participants in this burgeoning field range from graduate students supplementing their studies to parents seeking flexible income to full-time professionals who rely exclusively on this form of work. The scale of such efforts has given rise to a shadow workforce that, while critical to AI progress, remains susceptible to the same volatility observed in Uber’s Sandbox initiative.
For those with firsthand knowledge of Uber’s operations or experiences related to this story, Business Insider invited readers to reach out directly to reporter Alex Bitter using a personal email account at abitter@businessinsider.com or contacted securely through the Signal messaging app at 808-854-4501. To help ensure confidentiality, the publication recommends using a nonwork Wi-Fi network and device, following its published guidelines for secure information sharing.
Sourse: https://www.businessinsider.com/uber-project-sandbox-shows-ai-training-contractors-the-door-2025-11
