ホモグリフ

What is a Homoglyph?

A homoglyph is a character that looks visually identical or nearly identical to another character but has a completely different Unicode code point, name, and meaning. The word comes from the Greek homos (same) and glyphe (carving or symbol). Because modern typefaces render these characters with the same shape on screen, human eyes — and sometimes software — cannot distinguish between them.

The most well-known example is the Latin lowercase letter a (U+0061) and the Cyrillic lowercase letter а (U+0430). They are rendered identically in most fonts, yet they are entirely different code points belonging to different Unicode scripts. Dozens of such pairs exist across Latin, Greek, Cyrillic, Armenian, and many other scripts.

Why Homoglyphs Are a Security Problem

The Unicode Standard encodes characters from over 150 scripts, and many scripts independently developed symbols that resemble those in other scripts. This is expected and linguistically valid. The security problem arises when attackers deliberately substitute one character for another to trick users into believing they are looking at something they are not.

Common targets include:

• Domain names: The Internationalized Domain Names in Applications (IDNA) standard allows non-ASCII characters in domain names. An attacker can register pаypal.com using a Cyrillic а and create a convincing phishing site that appears to be paypal.com to a casual viewer.
• Usernames and handles: Social platforms that allow Unicode usernames are vulnerable to impersonation attacks where a fake account mimics a real one character-for-character.
• Source code and filenames: Homoglyphs in variable names or filenames can introduce subtle backdoors that are nearly impossible to spot during code review.

Common Homoglyph Pairs

Many scripts contribute characters that visually overlap with Latin letters:

• Latin o (U+006F), Cyrillic о (U+043E), Greek ο (U+03BF) — all look like "o"
• Latin p (U+0070) and Cyrillic р (U+0440) — identical lowercase forms
• Latin c (U+0063) and Cyrillic с (U+0441) — identical lowercase forms
• Latin e (U+0065) and Cyrillic е (U+0435) — identical lowercase forms
• Latin H (U+0048) and Cyrillic Н (U+041D) — identical uppercase forms

This means the word "COPE" written entirely in Cyrillic characters — СОРЕ — looks exactly like the Latin word "COPE" in most fonts.

How to Detect and Prevent Homoglyph Attacks

Unicode Technical Report #39 (Unicode Security Mechanisms) defines a confusables dataset that maps thousands of characters to their "safe" visual equivalents. Software can use this dataset to normalize or flag suspicious text.

Common defenses include:

1. Script mixing detection — reject or warn when a string contains characters from more than one script
2. Confusables normalization — map potentially confusing characters to a canonical form before storage or comparison
3. Punycode display — browsers display internationalized domain names in Punycode (xn--...) form when mixed scripts are detected
4. Visual diff tools — security-aware editors can highlight characters that are not in the expected script

Quick Facts