This chapter discusses the contentious world of web scraping and its essential, yet problematic, role as the lifeblood of generative AI (GenAI) model training. It highlights the vast scale of data required and the subsequent reliance on the internet’s reservoirs: the public, deep, and dark webs. The central tension explored is the conflict between technological capability and creator consent, revolving around legal and ethical dilemmas.

The chapter scrutinizes the ethical implications of this “finders keepers” approach, the use of torrenting to acquire copyrighted material, and the economic imbalance created when AI models compete with the original content creators. Finally, the chapter examines the complex debate surrounding content opt-out mechanisms and concludes by stressing the impending challenge of diminishing data, which will likely force a greater reliance on synthetic data. Ultimately, the chapter argues that the current approach has created a crisis of legitimacy that requires a reevaluation of the terms under which the internet should be scraped.