Digital labour means two different things depending on who is using it, and the two are close to opposites. One is a term from media and labour studies, in print since 2000, for the unpaid or underpaid work that sustains the digital economy. The other is Microsoft's, from April 2025, for AI agents bought on demand to expand a workforce. Neither owner is this estate, and the newer sense is not a coinage: it reuses two existing words without crediting the field that built them.
The answer, in one line
Digital labour has two established but incompatible senses. In media and labour studies, since 2000, it names the unpaid or underpaid work that sustains the digital economy: content moderation, data labelling, platform microwork, and ordinary social media use.
Definition#
Digital labour: in the older, academic sense, the work, much of it unpaid or underpaid, that produces value for a digital platform: content moderation, data labelling, platform microwork, and the ordinary use of social media that generates the data a platform sells. In Microsoft's 2025 sense, AI agents purchased on demand to scale workforce capacity. The two senses are not compatible, and a reader has to work out which one a source means.
Where the phrase was built#
Tiziana Terranova set out the underlying argument in 2000, in an essay called Free Labor: that a great deal of what keeps the digital economy running, from early online community-building to the content people produce for platforms, is labour that goes unrecognised as such because it feels voluntary and often enjoyable. Trebor Scholz gave the idea its current name at a conference held at The New School on 12 to 14 November 2009, titled The Internet as Playground and Factory: A Conference on Digital Labor, and in the 2012 book of essays that followed it. Christian Fuchs then gave the term its fullest theoretical treatment in Digital Labour and Karl Marx, 2014.
What the established term actually covers#
Fuchs describes an international division of digital labour: mineral extraction for the hardware, assembly-line manufacturing, software engineering, and the unpaid activity of a social media user, all contributing to one industry and all, in his argument, sites where capital extracts value. A content moderator reviewing violent material for low pay in an outsourced hub and a person scrolling a feed for free look nothing alike. What the term claims they share is their position relative to the platform: both produce value the platform captures, and only one of them is paid for it.
What Microsoft renamed in 2025#
Microsoft's 2025 Work Trend Index Annual Report, surveying 31,000 knowledge workers across 31 markets and published 23 April 2025, reports that "82% of leaders expect to use digital labor to expand their workforce in the next 12 to 18 months." The report frames this against what it calls the Capacity Gap: a stated mismatch between the productivity a business demands and the maximum output its people can supply unaided. AI agents, in this framing, are a form of labour a business buys to close that gap, alongside the people it employs.
Nothing in the report cites the older literature, and the two ideas are not reconciled anywhere in it. A phrase built to argue that uncompensated human activity deserves recognition as work has been reused to describe a line item a business purchases.
The same two words doing opposite jobs#
Calling something labour, in Terranova's and Fuchs's sense, is an argument: this activity produces value and the person doing it should be recognised, and usually paid, for it. Calling AI agents labour does something close to the reverse. It moves a software purchase out of the capital column and into the one a reader reserves for people, which is what turns "expand your workforce with digital labour" into a sentence that sounds like a staffing decision rather than a budget line. That is a framing choice, not a finding. Read it as one before repeating the phrase in a board paper or a job description.
Telling the two apart on sight#
A source using the older sense is usually writing about platforms, gig work, content moderation or data annotation, and is usually critical of how little of the value produced reaches the person producing it. A source using Microsoft's sense is usually a 2025-or-later document about scaling a workforce with AI agents, and is usually trying to sell something. Checking which one a document means, before adopting the phrase, is the only reliable move: the two cannot both be intended at once, and most writing that uses the phrase does not say which it has in mind.
Key sources
- Terranova, T. (2000). Free Labor: Producing Culture for the Digital Economy. Social Text, 18(2), 33-58.
- Scholz, T., ed. (2012). Digital Labor: The Internet as Playground and Factory. Routledge.
- Fuchs, C. (2014). Digital Labour and Karl Marx. Routledge.
- Microsoft and LinkedIn (2025). 2025 Work Trend Index Annual Report: The Year the Frontier Firm is Born. Microsoft WorkLab, 23 April 2025.
Related SuperSkills research#
On the term this one is paired with in Microsoft's own report, what is a frontier firm. On the boundary Microsoft draws around what a business keeps for people, what is human-reserved. On what happens to entry-level roles when agents take the work that used to train juniors, the missing rungs.
About this definition#
Digital labour is an established term from media and labour studies, corrected on this page from an earlier entry that had credited it to Microsoft. It is not a SuperSkills coinage, and Microsoft's 2025 usage is not a coinage either: it is a reuse of an existing phrase for an unrelated meaning, with no reference back to the literature it came from. Rahim Hirji is the author of SuperSkills (Kogan Page, 2026). Findings are attributed to the studies that produced them and kept separate from the interpretation.
Explainer · SS-2026-303 · Graded against the published rubric
Hirji, R. (2026). What is digital labour?. The SuperSkills evidence base, SS-2026-303. https://thesuperskills.com/research/what-is-digital-labour. Last reviewed 24 September 2026.
An evidence review by Rahim Hirji, not peer-reviewed research. For a material claim, cite the underlying study as well; every study here carries its own permanent link.
How citations and IDs work