Will cleaning make Cyrillic lookalike letters into Latin letters?

Measured inputs and outputs

These are locally constructed test strings, not evidence that Claude inserts these characters. We executed the saved homepage script snapshot with a minimal DOM harness and compared exact strings. The invisible-checkbox state and dash mode appear in each row. The observations below test this page's specific question. We make no detector-score or statistical-watermark removal claim.

The character mismatch remains visible in the data

The first fixture has U+0430 where the Latin reference has U+0061. Cleanup leaves the first string unchanged, and strict comparison with paypal stays false. The mixed fixture loses U+200B but keeps U+0430, so its strict comparison also remains false. The Latin reference stays equal. The example is a plain string without a domain suffix, destination link or network request. Its code-point list is the evidence; a similar glyph in a font is not evidence of character equality.

A text cleaner is not a confusable classifier

Unicode security guidance describes mechanisms for identifying confusable strings and script restrictions. We link that primary material but do not implement its skeleton mapping or test a general security detector. Cyrillic a is a legitimate letter in its own script. Transliteration or replacement based only on a Latin-looking word can damage intended multilingual text. This experiment establishes one explicit mismatch and shows that removing a separate selected marker leaves the mismatch intact. It makes no claim that a named brand is impersonated in an actual document.

Check identifiers against a known spelling

If an identifier or account name fails, compare it with the authoritative expected sequence and inspect each differing code point. Confirm the intended script before editing a working copy. For URLs or security-sensitive identifiers, use the destination platform validation and dedicated confusable protections. Do not accept a link simply because a character cleaner reports no findings. The homepage neither transliterates alphabets nor authenticates names, and a removal count cannot validate ownership or eliminate all spoofing risks.

Reproduce this test

Save reproduce.cjs and tested-app.js in the same folder. Run the command below with Node.js. The harness prints its runtime, script SHA-256 and every measured row. Compare those rows with the original record. Using a newer script or runtime creates a new experiment; retain the version information with your rerun.

node reproduce.cjs