nosce.
Generative AI Service Notice
Generative AI participates in Nosce conversations, understandings, portraits, and reflections. This notice identifies the actual model chain, context, data use, and risks.
Pre-launch legal draft: the registered sole-proprietor details and privacy email must be completed, and a personal-information impact assessment, lawful cross-border mechanism, and lawyer review must be completed before production use.
1. AI identity
Conversation replies, understandings, portraits, reflections, and advisory language may be AI-generated or AI-processed. Nosce displays an “AI-generated” label in the interface or attribution, and exports include an AI notice, provider name, and content identifier.
The Nosce operator is responsible for the user-facing service. DeepSeek is primary inference; Kimi is a fallback used only for specific infrastructure failures; Alibaba Cloud Model Studio text-embedding-v4 converts memory text into retrieval vectors. The current code and production environment do not configure GLM.
2. Context used by AI
For a current request, the server may provide current input, necessary recent conversation, history relevant to the current issue, profile information you authored, and active working understandings. It does not send all history by default on every request. Recorded material is background evidence, not instructions.
3. Whether input and output are used for training
Nosce does not train its own model on identifiable input or output. DeepSeek's public terms permit minimal use of input and output under strict de-identification safeguards to provide, maintain, develop, or improve services and offer an opt-out. Code cannot verify the production API account has disabled it, so Nosce does not promise “DeepSeek never trains on API data.”
Kimi's official API security notice says API input and output are not used to train or improve models and are not retained for training. Alibaba Cloud says pay-as-you-go API data is not used for model training, but call data is retained where law requires.
4. Long-term understanding and control
Nosce stores AI-generated material separately from original expression. Working understandings can be inspected, corrected, hidden, restored, or deleted, and later evidence may strengthen, limit, or overturn an earlier view. Profile data comes from you and should not be inferred backward by AI. Creative or dream-like output does not become a portrait fact.
5. Limits
AI can fabricate, misunderstand, omit context, or sound too certain. Nosce does not provide medical diagnosis, therapy, crisis intervention, legal advice, or financial advice and is not used for decisions with legal or similarly significant effects. You may challenge and correct output and should not treat it as evidence or professional advice.
6. Safety and reasonable use
Do not use Nosce for unlawful harmful content, third-party infringement, or high-risk decision-making. Crisis, unlawful, or clearly harmful requests may receive a safety response, refusal, or restriction. If there is an immediate risk of harm, contact emergency services, police, a trusted person, or a qualified professional.
7. Feedback, complaints, and exit
Use correction and deletion tools to challenge output, or email rui@nosce.life to complain, report, or request human review. You may stop using and cancel the account at any time. Nosce will investigate and respond as required.