← 研踪 PaperTrailarXiv:2611.23704 · v1
Localizing Text Anonymization for Trustworthy AI: Extending RAT-Bench to Malaysian Microdata and PII
首次提交 2026-11-18 09:53
版本更新 2026-11-18 09:53
雷达观测 2026-11-19 19:08
摘要
As governments and organizations adopt foundation models for public services and workplace productivity, sensitive citizen, employee, and administrative text may enter LLM workflows through inference, retrieval, fine-tuning, or local model development. We examine whether text anonymizers transfer to new deployment settings by extending RAT-Bench to Malaysia using local microdata, Malaysian PII formats, and culturally grounded transcripts in Malaysian English and Bahasa Malaysia. We evaluate NER- and LLM-based anonymizers using an LLM attacker that infers attributes from anonymized text, measuring both re-identification success and text utility. Across Malaysian English and Bahasa Malaysia, the two LLM anonymizers provide the strongest explicit privacy-utility trade-offs, reducing Easy/Hard re-identification risk to 23-29% while preserving BLEU scores of 0.77-0.94. Non-LLM tools show sharper failure modes, either preserving utility while leaving high residual risk or reducing risk through severe over-redaction, especially in Bahasa Malaysia. These findings suggest that anonymizers should be evaluated against deployment-relevant languages, identifier formats, and cultural contexts before being relied on in public-sector, workplace, or local AI pipelines.
作者备注:Accepted at the Trustworthy AI for Good (AI4GOOD) Workshop at ICML 2026
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-18 09:53 · 最新观测版本
Localizing Text Anonymization for Trustworthy AI: Extending RAT-Bench to Malaysian Microdata and PII
As governments and organizations adopt foundation models for public services and workplace productivity, sensitive citizen, employee, and administrative text may enter LLM workflows through inference, retrieval, fine-tuning, or local model development. We examine whether text anonymizers transfer to new deployment settings by extending RAT-Bench to Malaysia using local microdata, Malaysian PII formats, and culturally grounded transcripts in Malaysian English and Bahasa Malaysia. We evaluate NER- and LLM-based anonymizers using an LLM attacker that infers attributes from anonymized text, measuring both re-identification success and text utility. Across Malaysian English and Bahasa Malaysia, the two LLM anonymizers provide the strongest explicit privacy-utility trade-offs, reducing Easy/Hard re-identification risk to 23-29% while preserving BLEU scores of 0.77-0.94. Non-LLM tools show sharper failure modes, either preserving utility while leaving high residual risk or reducing risk through severe over-redaction, especially in Bahasa Malaysia. These findings suggest that anonymizers should be evaluated against deployment-relevant languages, identifier formats, and cultural contexts before being relied on in public-sector, workplace, or local AI pipelines.
查看此版本解读 →查看此版本原文 ↗