This is, to my knowledge, the first published per-font danger rate analysis for Unicode confusables. It suggests that font choice is a meaningful variable in confusable risk, one that confusables.txt does not account for.
Nature, Published online: 24 February 2026; doi:10.1038/s41586-026-10298-w
。WPS下载最新地址对此有专业解读
Что думаешь? Оцени!
Мир Российская Премьер-лига|19-й тур，推荐阅读搜狗输入法下载获取更多信息
A pair like Cyrillic ԁ (U+0501) and Latin d scores 0.781 mean SSIM across 18 fonts. That sounds moderate. But it is pixel-identical (SSIM 1.000) in eight of those fonts: Arial, Menlo, Cochin, Tahoma, Charter, Georgia, Baskerville, and Verdana. An attacker needs only one font to succeed. The exploitable risk is the max, not the mean.。雷电模拟器官方版本下载对此有专业解读
Every standard font that includes Cyrillic reuses the Latin glyph outlines. This is a deliberate font design decision, not a rendering quirk. No visual inspection can distinguish them.