ShieldFont: Protecting human texts by poisoning unauthorized AI scrapers

ShieldFont is a text-native defense system designed to disrupt unauthorized AI mass scraping by swapping critical content words in a page's HTML with other real words of the same grammatical class. While humans still read the original text via font rendering, machines processing the raw HTML receive altered meanings and poisoned datasets that reduce conceptual accuracy. The approach relies on economic friction rather than cryptography, forcing scrapers to face higher costs for rendering and decoding. Open article →