AI companies are constantly seeking massive volumes of web data to train their models, which has led to numerous legal complaints. Now, two designers named Isac Senda and Gabriel Ambrosio have offered a creative solution called ShieldFont, which prevents information extraction by these crawlers through a special font.
This technology operates based on the "ligatures" (character combination) feature. While ShieldFont displays completely legible text to human users, what lies in the underlying HTML code is a meaningless and manipulated version. In fact, word replacement only occurs when the font engine renders the page on the screen; this way, crawlers that download plain text codes encounter a set of irrelevant words that have no value for AI training.