arXiv：cs.CL· Muhammad Rafiullah Memon, Viet Vo, Wanlun Ma, Yang Xiang·· 5 小时前AI 评分35
儿童 AI 的手语转文本安全接口审计：Not What a Child Expressed
Not What a Child Expressed: Auditing the Sign-to-Text Safety Interface in Child-Facing AI
AI 导读
自动手语翻译（SLT）已进入消费产品，但尚无公开系统将 SLT 与面向儿童 AI 及平台信任安全工具联合评估，且已部署的主流 SLT 模型未在 18 岁以下手语者上训练或正式评估。翻译中改变否定、角色、保密、紧迫性或求助意图的错误可能在不影响流畅度的情况下改变安全判定。论文提出由聋人参与部署前审计该边界，含失败分类法、脱敏场景 schema、四种比较条件和四项结果指标，首例计划研究 Auslan。
正文
Abstract:Automatic sign language translation (SLT) has entered consumer products, turning American Sign Language into English text for dictation, messaging, and queries put to a conversational assistant. Child-facing AI and platform trust-and-safety tooling decide on text, using filters on minor accounts and grooming classifiers that score chat messages. A signing child who uses SLT therefore reaches these safeguards through a translation. We found no publicly documented system in which the two have been jointly evaluated, and the leading deployed SLT model was neither trained nor formally evaluated on signers under 18. Errors that alter negation, participant roles, secrecy, urgency or help-seeking could change a safety decision without disturbing fluency. This paper proposes a Deaf-informed pre-deployment audit of that boundary, with a failure taxonomy, a sanitised scenario schema, four comparison conditions, and four outcome measures. Auslan is the planned first case study.
| Comments: | 5 pages, 1 figure. Accepted as a poster at the NeurIPS 2026 Workshop on Child Safety in AI (non-archival) |
| Subjects: | Computation and Language (cs.CL); Cryptography and Security (cs.CR); Computers and Society (cs.CY) |
| Cite as: | arXiv:2610.07519 [cs.CL] |
| (or arXiv:2610.07519v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07519 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Muhammad Rafiullah Memon [view email]
[v1]
Mon, 5 Oct 2026 23:35:30 UTC (14 KB)
来源：arXiv：cs.CL · arxiv.org