Ping's approved architecture pairs 100% on-device speech recognition with prepared pre-voiced puppy audio. The puppy selects a prepared response rather than generating an open-ended cloud voice conversation.
Ping is a pre-launch product. This page describes the required zero-child-voice-recording design, not a claim that every historical build is compliant or that a public release has passed device acceptance.
Speech is recognized transiently on the device. No child audio recording is created, saved, sent to a cloud voice provider, retained, or used for model training. Puppy replies are prepared pre-voiced replies, not recordings of the child.
Where on-device recognition is unavailable, voice input must stay off. There is no permitted cloud-recognition fallback.
A parent controls permissions and age-appropriate access. Self-attesting to adulthood, receiving a gift, or signing into a website is not by itself verifiable parental consent under COPPA.
Voice privacy does not eliminate separate obligations for profile, consent, support, or technical data. No COPPA certification or blanket safety guarantee is claimed.
The pre-voiced design is intended for playful, supportive interaction and social-emotional learning activities. Recognition can misunderstand speech, and prepared responses cannot understand every situation. Ping is not therapy, a clinician, an emergency service, or a replacement for a trusted adult.
Parents should supervise, stop an uncomfortable interaction, and contact appropriate local emergency services for urgent danger. Do not put private child experiences into support messages.