The Web’s Newest Weapon Against AI Scrapers is a Font

The latest innovation in the battle against AI scrapers has come from an unexpected place: font design. ShieldFont, created by Isaque Seneda and Gabriel Abrucio, aims to “poison” AI training data without compromising readability for humans.

AI scrapers have become a significant problem on the web. These malicious programs scour websites for valuable training data, often without permission from owners. This has led to lawsuits and technical fixes aimed at stopping them. ShieldFont’s designers are taking a creative approach to this issue by leveraging font technology.

ShieldFont uses ligatures – a feature that replaces certain letter pairs with more readable versions when they’re close together – but instead of improving readability, it substitutes entire words with others. This substitution occurs only when the font engine draws the page on screen, leaving scrapers that download plaintext source code unaware of the change.

However, this solution is not foolproof. Sophisticated scrapers can easily bypass ShieldFont by downloading underlying HTML and re-rendering it themselves. This requires significant technical expertise, but it also means that AI scrapers will continually adapt to new security measures.

The implications of this are far-reaching. Widespread adoption of ShieldFont or similar tools could lead to a “font war” – an arms race between designers and developers as they try to outsmart each other’s fonts. This would be a waste of resources and raise important questions about the role of font design in web development.

Moreover, this solution acknowledges that machines can perceive the world differently than humans do. By creating a font that deceives AI scrapers without affecting human readers, we’re recognizing the capabilities and limitations of AI. This is a remarkable admission, one that speaks to the growing understanding of AI’s role on the web.

For web developers, this raises questions about whether they’ll need to design fonts specifically for AI resistance. How will this affect the broader web ecosystem? As we continue down this path, it’s worth asking whether we’re creating more problems than we’re solving.