A new poll reveals that 26% of American workers shelled out money for artificial intelligence services and then deployed them on the job within the past week. The survey, conducted by Epoch AI in partnership with Ipsos during March 2026, questioned more than 2,000 employed adults across the United States. The central finding: workers who pay for AI are far more likely to use it for actual work tasks than those relying on free versions.
The data reveals stark differences in how payment structures shape usage patterns. Half of all surveyed workers—50%—reported using AI tools in the preceding week for either professional or personal purposes. Among that group, half said their professional usage matched or exceeded personal use. The split becomes sharper when examining who foots the bill. Workers accessing AI through employer-provided subscriptions used the technology for work tasks 76% of the time. Self-funded subscribers applied AI to work assignments 58% of the time, while those sticking to free tiers put the tools toward professional tasks just 38% of the time. When asked about AI's impact on their daily responsibilities, 27% of work-focused users said the technology had taken over tasks they previously handled themselves, and another 21% reported that AI enabled them to tackle entirely new categories of assignments. These figures align roughly with separate tracking from Gallup, which found that frequent workplace AI use—defined as multiple times weekly or more—reached approximately 26% in late 2025. Gallup's measurements showed daily workplace deployment hovering around 12%, with usage at least several times annually landing between 45% and 52%. Across various 2026 surveys, estimates of generative AI use at work range from 21% to 45%, depending on question phrasing and timing.
The report notes a quieter consequence for companies: if 58% of self-paying subscribers are channeling AI toward work assignments, a portion of employees are effectively bankrolling their employer's productivity gains with personal funds. For individual workers, the task-replacement statistics warrant close attention—27% of work-oriented users said AI supplanted existing duties, while 21% said it unlocked new work types. The survey identifies the gap between occasional and daily adoption as the critical metric going forward, given that Gallup's daily figure of roughly 12% falls far below the approximately half of workers who touch AI at least a few times annually.
The data suggests employers face a hidden subsidy problem: staff members purchasing their own subscriptions are shouldering costs that might otherwise fall on the company balance sheet. Workers confronting the 27% task-replacement figure need to weigh whether AI is freeing them for higher-value assignments or simply eroding their role. The pattern to watch, according to the survey, is whether the 12% daily-use cohort expands to capture more of the 50% who engage with AI occasionally—a shift that would signal the technology moving from experiment to infrastructure. The bottom line: paying for AI and using it for work are tightly linked, and the workers opening their wallets are already integrating these tools into their daily routines at rates that dwarf free-tier adoption. Leaders hesitant to fund enterprise licenses may discover their teams are already solving that problem themselves, creating scattered subscriptions and fragmented workflows that bypass central oversight. The strategic question isn't whether AI enters the workplace, but whether companies will direct that entry or simply inherit whatever patchwork their employees assemble on their own dime.