Meta-Backed Scale AI's Outlier Platform Uses Gig Workers to Scrape Personal Data
Scale AI, a company 49%-controlled by Meta, has recruited experts from fields like medicine, physics, and economics through its Outlier platform to refine artificial intelligence systems. However, workers reveal that tasks often involve scraping personal data from social media, copyrighted artwork, and explicit content, diverging from the advertised high-level work.
Desperation and Ethical Dilemmas in AI Gig Work
Tens of thousands of individuals, including journalists, graduate students, and teachers, have been paid by Scale AI to train AI models. Many described desperation in an economy threatened by AI automation, with one worker stating, "A lot of us were really desperate." They expressed guilt over contributing to systems that could replace their own jobs, with an artist noting "internalised shame and guilt" for automating their dreams.
Scraping Social Media and Copyrighted Content
Workers reported scouring Instagram and Facebook accounts, tagging individuals by name, location, and friends, including minors. Tasks required ordering photos by age and harvesting copyrighted artwork to train AI in generating images. One assignment explicitly prohibited AI-generated images, instructing workers to use only human-created art, yet workers plumbed social media for new data as options dwindled.
Explicit and Unsettling Tasks
Taskers described transcribing pornographic soundtracks, labelling photos of dead animals or dog faeces, and handling violent police calls. A doctoral student recounted labelling a diagram of baby genitalia, despite assurances of no nudity or gore. Scale AI claims it shuts down inappropriate tasks and avoids child sexual abuse material, but workers felt uncomfortable with the content.
Monitoring and Pay Issues
Workers are monitored via Hubstaff, which screenshots visited websites, though Scale AI says it ensures accurate pay. Accusations of "bait-and-switch" tactics involve promising high salaries during recruitment, then offering lower-paid projects. Workers submit to unpaid AI interviews, believed to train AI, and face unstable employment with occasional mass layoffs.
Corporate Connections and Client Base
Scale AI's former CEO, Alexandr Wang, is Meta's chief AI officer, and its former managing director, Michael Kratsios, advises former US President Donald Trump. Clients include Google, Meta, OpenAI, the US Department of Defense, and Qatar's government. OpenAI ended its partnership in June 2025, citing ethical standards, while Meta and Anthropic did not comment.
Worker Uncertainty and Future Outlook
Taskers remain uncertain about AI's end use, with some speculating they trained Meta's Avocado model. Despite ethical concerns, many continue due to limited alternatives. One worker said, "I have to be positive about AI because the alternative is not great." Scale AI emphasizes flexible, project-based work with transparent pay, valued by skilled contributors.
