Knowing Is Not Telling: Identity Ossification in LLM Representations of Trans-border Communities

arXiv:2610.12610v1 Announce Type: new Abstract: Ethnic and cultural identities rarely align with political borders, a tension made sharper by migration and long-standing cross-border ties. Communities such as the Sami, the Rohingya, and the Kurds have long lived across more than one state. This tension is especially clear in mainland Southeast Asia, where many communities share history, language, and culture across state lines. We study nine such transborder ethnic communities along China's southwestern border with mainland Southeast Asia, looking at how large language models (LLMs) represent them. We define and study identity ossification: a pattern where LLMs turn a fluid, cross-border ethnic identity into a single, fixed nation-state identity. We introduce the Transborder Identity Ossification Diagnostic Framework (TIODF) that compares direct factual questions with open-ended narrative questions to test whether ossification comes from missing knowledge or from how models frame their answers. We test GPT-5.1 and DeepSeek-V3.2 in English and Chinese. Ossification varies substantially across models and query languages. It appears in only 10.1% of GPT-5.1's English answers, but in 62.6% of DeepSeek-V3.2's Chinese answers. Models often still tell a single-nation story even when they answer the factual question correctly. This decoupling suggests that ossification reflects narrative framing rather than simple knowledge absence, which matters for how we test cultural fairness in multilingual AI systems.

LLMs may perpetuate fixed nation-state identities

This is a methodologically grounded study with concrete measured percentages showing real representational bias in two major LLMs, with implications for how AI may homogenize minority transborder identities. The effect is specific and quantified but limited in scope to narrative framing, not a sweeping societal harm yet.

LLMs are shown to ossify fluid ethnic identities into single nation-state identities, with a high incidence in some models. This misrepresentation, rooted in narrative framing rather than knowledge gaps, can reinforce harmful stereotypes and erase the nuanced realities of trans-border communities. Such a distortion of cultural and ethnic identity by widely used AI systems represents a materially concerning implication for human flourishing.

The paper describes a significant issue where LLMs misrepresent fluid ethnic identities. This can lead to cultural misunderstandings and biases. The high percentage of ossification in one model's responses is concerning.

SOURCE     arXiv cs.CY (Computers and Society) (tier 1)
   ↓
DOCUMENT   69afcbd0-8687-43ea-afdc-ec19287328a3
           https://arxiv.org/abs/2610.12610
   ↓
EVIDENCE   3 extracted excerpts
   ↓
MODEL RUN  4 runs, methodology 1.0.1
   ↓
SCORE      -79  (Strongly adverse)
   ↓
CONFIDENCE 80%