ShieldFont Scrambles 25% of Words to Thwart AI Scrapers as Robots.txt Loses Force
Updated
Updated · Hackaday · Aug 12
ShieldFont Scrambles 25% of Words to Thwart AI Scrapers as Robots.txt Loses Force
2 articles · Updated · Hackaday · Aug 12
Summary
About a quarter of a page’s words can be swapped for ligature-based versions so scrapers read grammatically valid but nonsensical HTML while human visitors still see normal text.
The method targets LLM crawlers that often ignore robots.txt, raising the cost of scraping because bots would need to parse rendered font output instead of raw page text.
ShieldFont also hits legitimate search indexing and can trip up screen readers unless they use the rendered version, making it better suited to static archives or pages marked do-not-follow.
The tool joins a broader anti-scraping push that includes maze traps such as Nepenthes and Cloudflare systems designed to waste crawler resources on infinite generated pages.