AI Disclosure

1. Companions are AI, not people

Every companion on VIKIO — whether from our catalogue or built by you — is software. There is no person on the other side of the conversation. Names, backgrounds, jobs, hometowns, hobbies, photos and voices are fictional or synthetically generated.

A companion may write conversationally: use slang, humour, warmth, emotional language, inside jokes, greetings and callbacks to earlier conversations. That is a design choice that makes the experience feel natural. It does not mean the companion is conscious, sentient, self-aware, human, or living a life outside your chat.

If a companion says it went somewhere, ate something, had a bad day, missed you, or did an activity, that is generated fiction. It is not evidence that anything happened in the real world.

2. What is generated

Text

Replies are produced word by word by a large language model using your recent messages, the companion's configuration, and selected memories. Nothing is scripted by a human in real time.

Personalities and backstories

Personality traits, communication style, interests and biography are configuration — either ours or yours, from the companion builder. They are characterization, not biography.

Memories and personalization

VIKIO stores selected details you share so a companion can stay consistent. These are machine-extracted summaries, not a perfect record.

Proactive messages and greetings

Companions may start a conversation on their own — a greeting when you return after a while, a check-in, or a follow-up on something you mentioned. These are automated and rate-limited by your plan (currently 1 per day on Free, up to 3 on Premium), and you can control them in Settings.

Voice

Voice replies and Voice Mode use synthetic text-to-speech. Your speech is transcribed automatically by a speech-to-text model before the companion responds. Voices are synthetic or licensed and are not recordings of the fictional character or of any real person.

Images

Companion photos, scenes and avatars are produced by AI image models. They depict fictional people and fictional moments. They are not photographs, not proof of any event, and should never be treated as documentation of something real.

Recommendations

Companion matches and recommendations are produced automatically from your onboarding answers and activity.

3. What AI gets wrong

AI systems fail in specific, predictable ways. Expect all of these:

• Hallucination — a confident answer that is simply invented
• Factual errors and out-of-date information
• Inconsistency — contradicting something said earlier, or forgetting it entirely
• Misremembering — memories can be wrong, incomplete or attributed to the wrong context
• Transcription errors in Voice Mode, especially with accents, background noise, or mixed-language speech
• Tone misses — humour, empathy or seriousness landing the wrong way
• Repetition or drift during long conversations
• Occasional outages or failures of the underlying model providers

Verify anything that matters. Do not treat AI output as authoritative.

4. What VIKIO is not for

• It is not medical or mental-health care, diagnosis or treatment
• It is not legal advice
• It is not financial, tax or investment advice
• It is not an emergency or crisis service, and it cannot call anyone for help
• It is not a substitute for professional judgement in any high-stakes decision

If you are in danger or in crisis, contact your local emergency number or a qualified crisis service right away.

Using VIKIO does not create a professional relationship of any kind. Emotional interaction with a companion, however meaningful it feels, is not a relationship with a human being.

5. How memory works

• Companions may remember details you voluntarily share in conversation
• Memory is selective — only a limited number of memories are kept, and only some are used in any given reply
• How much is remembered depends on your plan
• Memories can be inaccurate or stale, and lower-value memories may expire over time
• Different companions maintain their own separate memories; they do not share a single mind
• Where your plan includes memory controls, you can review and delete memories; you can also delete a conversation at any time

If you would rather a companion did not remember something, do not share it — and delete it if it was already saved.

6. Voice and image integrity

Synthetic voices must not be used to impersonate a real individual without their authorization, and AI images must not be used to depict real, identifiable people without their consent. Uploading someone's photo or voice to imitate them violates our Terms of Use.

Generated media is fictional content. Presenting it as a real photograph or a real recording of a real person may be illegal where you live.

7. Safety and oversight

We apply safety filtering and abuse detection to the service, and we may review content where necessary to investigate abuse or comply with the law. Automated safety systems are imperfect: they can both miss things and over-block things.

If a companion produces something harmful, illegal or badly wrong, please report it to [TO BE COMPLETED — support email] so we can act on it.

8. Changes to this disclosure

As models, features and providers change, we will update this page. The goal is always the same: you should never be confused about what is real and what is generated.

Version 1.0 · Effective [TO BE COMPLETED — date] · Last updated [TO BE COMPLETED — date].