Some journalists are taking freelance jobs with AI training data companies like Scale AI, which recruit them for tasks such as fact-checking and prompt drafting
The gig work platform Outlier is one of several companies courting journalists to train large language models (LLMs). Bluesky: @andrewdeck , @tylerborchers.com , @tonymartin , and @amywestervelt . Mastodon: @aulia@mementomori.social See also Mediagazer Bluesky: Andrew Deck / @andrewdeck : Outlier is one of several gig work platforms recruiting journalists to train AI models. Its clients including Meta, OpenAI, and Microsoft. — I spoke to journalists working with the company for @niemanlab.org. Many are freelancers struggling to find full-time work. www.niemanlab.org/2025/02/meet... Tyler Borchers / @tylerborchers.com : “Their work indexes heavily on fact-checking, including identifying hallucinations by models or marking when chatbots pull from incorrect sources on the internet. Many of them compared it to ‘spot checking’ a story.” Tony Martin / @tonymartin : Meet the journalists sawing off the branch they're sitting on. www.niemanlab.org/2025/02/meet... Amy Westervelt / @amywestervelt : Anyone who does this is a fucking scab — www.niemanlab.org/2025/02/meet... Mastodon: Aulia Masna / @aulia@mementomori.social : @Techmeme Outlier is just one of several platforms that offer journalists around the world to train AI in several different languages. I get a lot of these open roles on LinkedIn's job suggestions asking for AI trainers in my language and I know people who've taken it up to keep themselves afloat in a scarce job market. … See also Mediagazer
Context & Ripple Effects
This report extends an established shift in AI training labor: Scale AI and Appen had already recruited writers and poets for language-model improvement, while another account described a writer producing model responses for a tech employer. Journalists bring a more specific mix of verification, sourcing, and editorial judgment to that labor pool.
It matters because the work is being offered to freelancers facing scarce full-time newsroom employment, linking journalism’s precarious labor market to the human-feedback operations behind major AI customers.
First-order effects
- Journalist freelancers gain a new, if contingent, source of paid work through Outlier and comparable platforms.
- Outlier and Scale AI can apply reporters’ fact-checking, hallucination detection, source review, and prompt-writing skills to LLM training services for clients including Meta, OpenAI, and Microsoft.
Second-order effects
- AI-training platforms have a stronger incentive to recruit specialized editorial workers, rather than treating all annotation as interchangeable gig labor.
- The arrangement sharpens a conflict within journalism: work that supports AI systems can provide income to freelancers while drawing criticism from peers concerned about its effect on the profession.
Third-order effects
- If specialized knowledge work continues to be broken into platform tasks, editorial expertise may increasingly be purchased as on-demand model-improvement input rather than supported through durable newsroom roles.
- The pattern points to a broader contest over how human judgment is valued in AI supply chains—whether as a distinct professional service or standardized gig work.
The trend: AI developers and their intermediaries are institutionalizing human-feedback pipelines by drawing specialized workers, including journalists, into model training.