The Future of Unicode: What Comes After 16.0?

Unicode 16.0, released in September 2024, brought the total assigned codepoint count past 155,000 characters, spread across 168 scripts. With 1,114,112 total codepoints in the Unicode codespace, roughly 86% of that space remains unassigned. The standard is nowhere near full, and active work continues on many fronts — from ancient scripts that have never been encoded, to evolving technical algorithms, to Unicode's role in the age of large language models.

The Remaining Codepoint Budget

The Unicode codespace spans U+0000 through U+10FFFF — exactly 1,114,112 positions. That number is determined by the UTF-16 encoding's surrogate pair mechanism, which limits code points to 21 bits. As of Unicode 16.0, the breakdown looks approximately like this:

The large reserved space means Unicode has room to grow for a very long time at current rates of addition. The bottleneck is not codepoint scarcity but rather the research, expert consultation, and committee review required to correctly encode each new script.

Scripts in the Pipeline

The Unicode Technical Committee (UTC) maintains a public roadmap of scripts under consideration for future versions. Some notable candidates:

Khitan: The Khitan people ruled the Liao dynasty (907–1125 CE) and used two distinct scripts — Khitan Large Script and Khitan Small Script. Khitan Small Script was added in Unicode 13.0. Khitan Large Script encoding is still under development due to the complexity of identifying and cataloguing the characters from historical artifacts.

Proto-Elamite: One of the world's oldest writing systems (~3200–2900 BCE), used in ancient Iran, remains largely undeciphered. Encoding undeciphered scripts is a deliberate Unicode goal — it allows scholars to digitize and search historical documents even without knowing the script's meaning.

Jurchen: The script used by the Jurchen people (12th–17th century), ancestors of the Manchu. Encoding efforts are ongoing.

Toto: A small script used by the Toto people of West Bengal and Bhutan, added in Unicode 14.0 as an example of encoding living minority scripts.

Sutton SignWriting: A system for writing signed languages visually. Already in Unicode (15.0), but expanded coverage is planned.

The pattern is clear: Unicode continues expanding both toward ancient/undeciphered scripts (historical preservation) and living minority scripts (digital inclusion).

Emoji Evolution

Emoji are added to Unicode annually, with the Emoji Subcommittee evaluating hundreds of proposals. The trends shaping emoji's future:

Personalization and combination: The existing skin tone modifier system expanded to multi-person emoji (couples, handshakes) with independent tone selection. Proposals exist for hair color modifiers beyond the current hair component system.

Animated emoji: Unicode itself defines static codepoints; animation is entirely a platform-level concern. However, platform coordination (via the Emoji Requests process) increasingly considers animatability as a factor. Apple, Google, and Meta all animate emoji in their implementations; the question is whether any cross-platform animation standard will emerge.

Narrowing new emoji additions: After rapid growth in the 2015–2020 period, the criteria for new emoji have tightened. The UTC now requires demonstrated high usage potential, distinct semantic value, and global rather than regional relevance. The era of hundreds of new emoji per year is likely ending.

Technical Algorithm Improvements

Beyond new characters, Unicode's technical algorithms continue to evolve:

Line Breaking (UAX #14): Line breaking rules are among the most complex in Unicode, with ongoing edge cases discovered as new scripts are encoded and as web rendering engines report mismatches. Each Unicode version typically includes refinements.

Bidi Algorithm (UBA, UAX #9): The Bidirectional Algorithm, governing how left-to-right and right-to-left text are mixed, has been updated repeatedly to close security gaps — notably the "Trojan Source" class of attacks using bidi override characters to mislead human code readers.

Identifier Security (UTS #39): As programming languages adopt Unicode identifiers, the security considerations multiply. UTS #39 specifies confusables (characters that look similar to each other) and recommends restrictions. Updates track newly discovered homograph attack patterns.

Segmentation (UAX #29): Word and grapheme cluster boundaries need updating as new scripts are added and as edge cases in existing scripts (particularly for Indic scripts with complex conjunct behavior) are resolved.

Unicode's Role in AI and NLP

Large language models tokenize text using algorithms (BPE, Unigram, WordPiece) that operate on Unicode strings. Several Unicode-level factors affect AI systems:

Normalization consistency: Models trained on unnormalized text may learn different representations for canonically equivalent strings. Production NLP pipelines typically apply NFC normalization at input boundaries.

Tokenizer handling of Unicode: Tokenizers trained primarily on English and Latin-script text often tokenize non-Latin scripts inefficiently — a single Arabic or Devanagari word may consume many tokens, disadvantaging speakers of those languages in effective context window size. This is a recognized problem the AI research community is actively working on.

Unicode security in AI outputs: LLMs can generate bidi override characters, zero-width characters, or homoglyphs in ways that create security risks if outputs are rendered in security-sensitive contexts (code, terminals). Sanitizing LLM outputs for dangerous Unicode sequences is an emerging concern.

Low-resource script support: Many of the world's scripts are drastically underrepresented in training data. Unicode's continued expansion of script coverage is a precondition for multilingual AI equity.

The Long View

Unicode has been called one of the most successful infrastructure projects in computing history. It unified a chaotic landscape of incompatible encodings into a single coherent system that now underlies virtually all digital text. The remaining work — encoding the last undocumented scripts, refining algorithms for edge cases, adapting to new security threats, and integrating with AI systems — is less dramatic than the original unification project but equally important.

With 86% of its codepoint space still available and an active technical committee, Unicode is well-positioned to handle whatever the next century of human communication requires.