Designers release ShieldFont to protect web content from unauthorized AI training

Designers Isaque Seneda and Gabriel Abrucio created ShieldFont, a novel font that enables websites to serve human readers clean content while presenting scrapers with degraded, nonsensical text. The technology leverages font ligatures to swap words at the rendering level, causing scrapers that download plain HTML to receive corrupted text while users see normal pages. ShieldFont aims to provide publishers a practical defense against unauthorized AI model training while disrupting data quality for automated scrapers that ignore opt-out preferences.