AI proposes. Authority stays explicit.

AI transparency

Users should be able to tell when they are interacting with AI. Model-generated content can be useful and still be wrong. Material decisions require the appropriate evidence and authority rather than confidence alone.

Provider independence

Where multiple model providers are available, routing is treated as an implementation capability. Provider and model provenance should be retained where materially relevant. Provider substitution does not transfer identity, safety, consent or action authority to the provider.

Personal information in AI

Input and generated output can both contain personal information. Data sent to an AI provider should be limited to what the requested feature needs and subject to the available privacy, security and contractual controls.

Training and system improvement

VexlorGrid does not use raw private conversations or identifiable Person history as unrestricted system-training material by default. Permissioned derived evidence is the preferred learning interface. A materially different training use requires a defined purpose, notice and any consent required for the data involved.

Automated decisions

The system is designed so consequential external actions and decisions do not gain authority merely because an AI recommended them. Where automated processing may significantly affect rights or interests, additional human, legal and transparency gates apply.