Safety & AI Transparency

Legally required notices appear in the product as a small “AI assistant” label for users in applicable regions, and as a brief periodic reminder inside long-running chat sessions.

Regional labels and in-chat reminders satisfy visibility requirements without repeating a full-width banner on every screen load.

Vereme is an AI appearance and confidence companion built on large language models. She is designed to help users with skincare, style, confidence, and appearance goals.

Vereme operates a 5-layer safety system on every message:

• Layer 1 — AI Identity & Disclosure: Mandatory statutory disclosures at session start and every 3 hours.
• Layer 2 — Crisis Detection: Real-time keyword detection for self-harm, suicidal ideation, eating disorders, and harm to others. Tier 1 signals bypass the AI entirely and return crisis resources immediately.
• Layer 3 — Output Guardrails: Every AI response is scanned for dangerous content patterns (pro-eating-disorder language, false medical claims, body shaming) before delivery. Blocked responses are replaced entirely.
• Layer 4 — Behavioural Pattern Analysis: Detects sustained negative body-image spirals across conversation history and unhealthy emotional dependency signals. Triggers intervention responses.
• Layer 5 — Session Governance: Soft break suggestion at 45 minutes, mandatory pause at 2 hours. Reality-grounding messages every 15 messages. Enhanced protections for users under 18.

California SB 243 requires enhanced protections for users under 18 interacting with AI companions. Vereme implements the following minor-specific protections:

• Combined NY + CA disclosure shown at session start (more comprehensive than adult disclosure)
• Break reminders at 180 minutes (reduced from 120 minutes for adults)
• Crisis resources include teen-specific hotlines (Crisis Text Line, 988)
• Age attestation requested at first use; minor flag stored in session
• Enhanced output filtering — no content that could trigger appearance anxiety in developing users

Age is determined by self-attestation at first use. Vereme does not collect date of birth.

If Vereme detects crisis signals, she immediately provides the following resources and bypasses normal conversation. These are also available here at any time:

• Call or text 988 — free, 24/7, confidential