The Unicode 18.0 draft adds 13,007 characters and changes line breaking, segmentation and spoofing rules

Sources: primary, enumerated versions, discussion
Summary: The 18.0.0 version page, dated 2026-09-16, carries a STATUS line reading that this is a preliminary draft page for an upcoming release, with further beta-review notices over the Annexes, the UCD data files and the migration section, and it states the version adds 13,007 characters for a repertoire of 172,808. The annex changes are the upgrade work: UAX #14 changes rule LB12a and moves FIGURE DASH, EN DASH and SOFT HYPHEN between line-break classes, UAX #29 revises grapheme cluster rule GB9c, and UTS #39 tightens spoofing rule A1 by removing the allowance for transparent characters between ZWNJ and the following joining character. The enumerated-versions table lists 18.0.0 as a 2026 release alongside every prior released version, which the draft notices on the version page contradict.
Why it matters: Existing text wraps, segments and gets flagged differently once a library picks up the new data, without any change to the text itself.
Follow-up: Check that the 18.0.0 version page drops its preliminary-draft and beta-review notices, and read the release announcement post on blog.unicode.org.