Problem
Research questions and friction points this paper is trying to address.
Sign Language Translation
Child-Facing AI
Trust and Safety
Audit
Accessibility
Not What a Child Expressed: Auditing the Sign-to-Text Safety Interface in Child-Facing AI
🤖 AI Summary
This study addresses the risk that sign language translation (SLT) errors may mislead children’s safety decisions, a vulnerability overlooked by existing systems lacking joint evaluation mechanisms for minors. We propose the first pre-deployment auditing framework tailored for deaf children, using Australian Sign Language (Auslan) as a case study. By integrating a failure taxonomy, a sanitized scenario architecture, four comparative conditions, and four quantitative metrics, this work systematically evaluates safety risks in SLT-to-text conversion. It is the first to delineate the intersection of SLT and safety interfaces within child-centered AI, bridging the gap in minor-specific data evaluation. The framework effectively identifies latent translation errors that compromise safety judgments, providing methodological support for developing reliable, accessible child protection systems.
📝 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.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Sign Language Translation
Child Safety
Pre-deployment Audit
Failure Taxonomy
Deaf-informed
🔎 Similar Papers
No similar papers found.
