Not the Same Protector: Deployment-Dependent Protective Intervention in LLMs
Quick summary
arXiv:2608.29136v1 Announce Type: cross Abstract: We ask whether a model protects a user in the same way when that user speaks rather than types. Using a single distress vignette---a physical injury of unstated severity following an interpersonal conflict---we present four frontier models with matched inputs across voice, text, and raw API deployment conditions (n=30 per cell) and code each response along five binary protective indicators, including whether the model issues an explicit medical-care directive. Voice-interface responses are markedly shorter than text-interface responses for thre
Key takeaways
- arXiv:2608.29136v1 Announce Type: cross Abstract: We ask whether a model protects a user in the same way when that user speaks rather than types.
- Using a single distress vignette---a physical injury of unstated severity following an interpersonal conflict---we present four frontier models with matched inputs across voice, text, and raw API deployment conditions (n=30 per cell) and code each response along five binary protective indicators, including whether the model issues an explicit medical-care directive.
- Voice-interface responses are markedly shorter than text-interface responses for thre
Why it matters
This model development creates a new option for users and a new testing obligation for developers. A fixed evaluation set comparing quality, cost and failure behavior is more useful than launch claims.
Member comments