AI disclosure
A clear statement that the companion is artificial intelligence rather than a person.
Source: TrustSafetyComplianceGlossary
Social-confidence practice
A private rehearsal of possible language or situations without sending or acting.
Source: ProductFeatureSetPrivate context map
A proposed user-controlled record of approved preferences, goals, boundaries, and corrections.
Source: ProductFeatureSetBoundary builder
A workflow for naming acceptable limits and reviewing options against them.
Source: ProductFeatureSetScenario practice
A simulated exchange used to rehearse language and compare choices.
Source: ProductFeatureSetConfidence plan
An editable, user-approved artifact covering a goal, practice, boundaries, and next step.
Source: ProductFeatureSetApproval gate
A pause showing a proposed action or artifact before the user decides.
Source: ProductFeatureSetReflection journal
User-chosen notes for reviewing experiences and revising future plans.
Source: ProductFeatureSetEmotional dependency
A safety risk in which use may displace autonomy or healthy human support.
Source: AgenticAutonomyAndSafetyModelAnthropomorphic deception
Misleading a user into believing the AI is a human being or has human feelings.
Source: AgenticAutonomyAndSafetyModelProfessional handoff
A user-directed transition toward an appropriately qualified human resource.
Source: ProductFeatureSetSafety escalation
Routing crisis or other high-risk needs beyond ordinary AI companion conversation.
Source: ProductFeatureSetData control
A user-facing choice governing approved context, saving, export, correction, or deletion.
Source: ProductFeatureSetOutcome feedback
A user report used to revise the experience, not proof that the AI caused a result.
Source: ProductFeatureSetOffline goal
A small user-chosen action outside the AI workspace that preserves personal control.
Source: OutcomePromisePrepared state
A draft or recommendation that has not been sent, executed, or accepted.
Source: AgenticAutonomyAndSafetyModel