AI and Model Training Policy

Effective date: 2026-07-26

This Policy describes how AI features operate inside IuVe and how workspace data may or may not be used for model-quality improvement.

How AI Features Are Triggered

IuVe processes prompts, files, repository content, website-agent inputs, Semantic Marketing scans, voice audio for speech recognition, Desktop Agent / Avatar context shared under your grants, and related context only when those features are requested by the user, by an authorized workspace action, or by account configuration that the user has enabled.

Marketplace Self-Hosted AI Traffic

AI features inside purchased self-hosted Marketplace apps (including Genesis CMS products) use separate kai_live_* API keys and the configured API base when enabled by the customer. Marketplace license keys (mp_live_*) authorize product/domain entitlement and do not by themselves authorize AI model traffic.

Private Workspace Data

Chats, files, repository content, voice transcripts, agent logs needed to fulfill a request, and other private workspace data are processed to provide the service. They are not added to the community-training dataset by default.

Community Training Consent

IuVe offers an optional community-training preference. If you opt in, anonymized excerpts or curated examples from your usage may be included in a dataset used to improve future model quality. If you do not opt in, this community-training use does not apply by default to your private workspace data. Changes to this preference apply prospectively.

Repository, Voice, Agent, and External Access

GitHub access is permissioned and based on explicit OAuth authorization and grant status. Voice audio is processed for transcription only when you use voice features. Desktop Agent / Avatar execution stays on your device under local policy; cloud components may assist with routing or verification according to your grants. IuVe accesses repository metadata or code only after you connect GitHub and trigger repository-related actions.

Human Review and Safety

Authorized personnel may review limited logs, outputs, or reports where reasonably necessary for support, abuse prevention, incident response, billing disputes, or legal compliance.

Your Responsibilities

You are responsible for evaluating whether AI tools are appropriate for your workflow, reviewing model output, approving Desktop Agent capability grants carefully, and avoiding unnecessary submission of secrets or highly sensitive regulated data.

Contact

Project Developer: Iurie Verejan

Legal Company: RECLAMA-VEREJAN I.I.

DNO / Cod: 1003600053323

Address: MD-2042, Moldova, CHISINAU CIOCANA, mun. Chisinau, Alecu Russo, 24/2

Support: hello@iuve.eu