AI is disclosed
Onboarding, policies, and in-product reminders identify every persona as AI. A persona may speak from a fictional biography inside chat, but the product does not present that identity as a real person.
Back to home
Trust Center
What Samagama stores, protects, discloses, and interrupts.
This is the plain-language map of controls currently built into Samagama: AI disclosure, field encryption, memory consent, export and deletion requests, usage limits, and crisis routing. The linked policies contain the full legal terms.
Principles
Controls you can verify in the product
Memory consent has a switch
The Privacy center records whether a user permits long-term conversation memory. Long-term memory extraction is not currently enabled; the same screen provides data-export and account-deletion requests.
Sensitive fields are encrypted
Designated sensitive fields are encrypted with AES-256-GCM before database writes. Row-level access rules restrict user-owned records to the authenticated owner or authorized server paths.
The product interrupts overuse
Long-session, high-volume, and late-night activity can trigger wellbeing nudges. Continuous sessions also have a mandatory-break control rather than an unlimited engagement loop.
Privacy
What the service handles
Samagama handles account details, discovery responses, conversation content, persona interactions, subscription state, safety events, and the minimum operational data needed to provide the service.
Conversation content and personality answers are not public profile material, are not sold, and are excluded from optional product-analytics events.
- Sensitive application fields use authenticated AES-256-GCM encryption before storage.
- Database row-level security separates one user's records from another user's records.
- Optional analytics run only after consent and do not receive chats, personality answers, or card details.
- Payment providers process card information; Samagama does not store full payment-card numbers.
Memory
How memory consent works today
The Privacy center stores whether a user permits long-term conversation memory. Long-term memory extraction is not currently enabled, so this setting records consent rather than activating a separate extracted-memory system.
Messages still remain available to run the active conversation and meet account, safety, and legal requirements. Conversation data is scoped to the user's experience, not published as a biography or shared as another user's persona context.
- Long-term conversation-memory permission can be paused from the in-product Privacy center.
- The same center exposes personality-discovery consent and request status.
- Authenticated users can queue a copy of their data or an account-deletion request.
- Personas are instructed not to ask for credentials, exact addresses, payment details, or government IDs.
AI transparency
Fictional identity without a fake human claim
A Samagama persona has a name, city, work history, schedule, interests, and a consistent speaking style. Those details create a coherent character; they are not evidence of a human being behind the account.
The product discloses AI during onboarding and through ongoing interface reminders. Inside a conversation, the persona may stay in character rather than discussing model architecture.
- Public persona pages describe the fictional identity before a user starts chatting.
- AI disclosure remains a product-level responsibility, not something hidden in legal text.
- Personas cannot take real-world actions or make account, billing, or privacy decisions.
- Generated replies can be wrong; important medical, legal, financial, or safety decisions require qualified human help.
Safety
What happens when use becomes intense
Samagama tracks session duration and usage volume so the interface can suggest a pause during long, repeated, or late-night use. A continuous-session limit can require a break instead of allowing the conversation to continue indefinitely.
If language suggests immediate danger or self-harm, the response is directed toward local emergency services, a crisis line, and a trusted person nearby. Samagama is not an emergency or clinical service.
- Safety instructions prohibit harmful guidance, coercion, exclusivity, jealousy, and encouragement to isolate.
- Wellbeing nudges cover long sessions, high daily use, and late-night activity.
- The prompt asks personas to encourage one real-world support step when a user says the persona is their only support.
- The service is for adults and does not allow sexual content involving minors.
Policy map
Where to read the formal details
The Trust Center is written for clarity. These linked pages contain the fuller policy language for privacy, AI behavior, acceptable use, and account terms.
Privacy Policy
The formal details on data collection, use, retention, export, and deletion.
AI Policy
How samagama.ai uses AI, what personas are, and what limits apply.
Usage Policy
The behavior rules that keep conversations safer and more respectful.
Terms of Service
The legal agreement for using samagama.ai.
Questions
Common trust questions
Are samagama.ai personas real people?
No. They are AI characters with fictional identities, voices, schedules, and boundaries. Onboarding and in-product reminders disclose that clearly.
Is conversation data public?
No. Conversation content is private product data, not public profile material.
Can users delete their account data?
Yes. An authenticated user can queue an account-deletion request from the Privacy center. A data-export request is available there as a separate action.
Is samagama.ai a substitute for urgent support?
No. AI companionship is not a replacement for urgent support, trusted people nearby, emergency services, or qualified professional care.
Need a specific answer?
Ask about privacy, safety, or your data.
The policy pages explain the formal rules. For an account-specific request or a question not covered there, use the contact page and choose the privacy or support route.