DRAFT — proposed notice for a feature that is NOT ACTIVE. Do not label current conversations “AI protected” or display a monitoring indicator before implementation and validation.
What safety assistance is intended to do
PPLONLY plans to use AI as an aid for identifying specific threats, targeted harassment, sexual coercion, exploitation, scams and privacy violations. It is not intended to rate personality, accent, emotion, politeness, attractiveness or political views. It cannot guarantee a conversation stays professional, and its outputs can be wrong. Human judgment, report controls and the ability to leave remain essential.
Before any live processing begins
The pre-call notice must identify: To be confirmed: ai provider and model, To be confirmed: audio video or text inputs, To be confirmed: live processing location, To be confirmed: transcript or buffer behavior, To be confirmed: automatic actions, To be confirmed: moderation retention, To be confirmed: human access, and To be confirmed: choice if declined. It must separately explain any recording. “Not recorded” must not be used to conceal transcription, temporary capture, logging or retained excerpts.
Both participants must receive clear notice and any legally required consent process before capture begins. Permission to use the microphone for a call is not permission for AI analysis. Silence, a preselected checkbox or agreement to general Terms is not a substitute for required specific consent. If the service requires monitoring for a particular offering, disclose that before payment and offer the approved non-monitored alternative or cancellation/refund when a person declines; do not surprise them after payment.
During and after a monitored session
When enabled, the interface must show the actual processing state and distinguish AI assistance, human review and recording. A reminder or temporary protective action is not a finding of guilt. Human reviewers must assess context before permanent sanctions or significant financial decisions, with a way to appeal. If monitoring is unavailable, tell participants and follow the announced fallback; never imply protection continues silently.
Recording and reuse
Platform recording is currently disabled. Future recording requires a separate purpose, an appropriate participant permission process, a persistent indicator, access controls, retention/deletion rules and a clear stop control. Moderation consent does not permit creating a shareable replay. Sharing, promotional use or model training needs its own lawful authorization. Private content is not approved for training general-purpose models by this draft.
Users may end a conversation or report concerns without waiting for AI. Questions and appeals: To be confirmed: safety contact. Privacy/withdrawal questions: cayla@pplonly.com. A withdrawal cannot undo already-lawful processing or erase evidence subject to a valid retention obligation; explain the applicable consequence at the time of the request.