Trend Insight 19: scarcity of quality data turns acquisition into a contest over rules

Summary

This trend describes the scramble for training data turning into a fight over the rules of access. Heavy investment in AI pushes companies to build models faster, which demands high-quality data that is growing scarce. Some social media platforms have rewritten their terms of service so user posts can train AI, and some firms scrape websites without permission even when content is marked off limits. In response, websites and creators are changing their terms, filing copyright lawsuits and using tools that block scraping or poison data, and one audit found 45% of a Google training dataset restricted by site terms. Companies are also turning to synthetic data and new scientific datasets, while the value of real data and the market for labelling it keep rising.