AI Giants Scrape Your Photos, Books, and Even Porn to Train Their Models

AI Giants Scrape Your Photos, Books, and Even Porn to Train Their Models A major artificial intelligence company, part-owned by social media giant Meta, is paying tens of thousands of gig workers to harvest personal, copyrighted, and explicit content from the internet to train its AI systems, according to worker accounts and company practices [122801]. Scale AI, which is 49% controlled by Meta, recruits credentialed professionals through its Outlier platform for flexible work training AI models. However, the reality of the job involves desperate measures, with workers tasked with scraping public Instagram profiles, transcribing audio from adult videos, and using copyrighted books and articles—often without the creators' knowledge or consent [122801]. This practice forms the hidden backbone of the AI boom, raising serious ethical and legal questions about how the world's most advanced AI systems are built. The data collection relies on vast amounts of personal and proprietary information to refine the models used by Meta and other technology giants [122801]. The scramble for training data comes as companies face an explosion of AI-generated computer code, a problem dubbed "AI tech debt," which is creating a massive backlog of messy, unmanageable software that human engineers must clean up [122144]. Simultaneously, skilled professionals, unable to find traditional employment, are turning to this low-paid, project-based AI training work as a critical last resort, highlighting a brutal shift in the job market [122822]. The UN has warned that the global development of AI, dominated by corporate and state powers outside the continent, risks a new form of "digital colonisation" in regions like Africa, where imported systems may not reflect local needs and could deepen global inequality [121720]. Meta's AI Trained by Gig Workers Scraping Social Media, Copyrighted Work, and Porn Desperate for Work, Skilled Professionals Turn to AI Training AI Code Explosion: Companies Can't Keep Up UN Scientist Warns of AI 'Digital Colonisation' in Africa