Assistive, bounded, reviewable
AI Data and Content Policy
A draft policy for AI usage metadata, provider governance, privacy, and human review.
AI data minimization
The current AI usage model stores request metadata for quota, reliability, and billing-related accounting. Raw prompts and model responses are intentionally not stored by default. Provider adapters must preserve this boundary unless a separately reviewed feature requires otherwise.
User guidance
Users should not submit passwords, session tokens, payment secrets, confidential school records, sensitive personal data, or material they lack permission to process. AI output may be incorrect, biased, incomplete, or unsuitable for a learner without review.
Providers and training use
Before a real provider is activated, Acadanex must document provider processing, geographic handling, retention, training/data-use settings, subprocessors, deletion controls, and contractual safeguards. No live provider is active in this phase.
Minors, files, OCR, and educational integrity
AI interactions involving minors, uploaded files, OCR, private document assistance, and assessment content require minimization, access controls, retention rules, human review where appropriate, and jurisdiction-specific legal review. AI assistance does not establish authorship or academic correctness.
Generated educational content
AI may assist with research, drafting, classification, metadata, editing suggestions, and question generation. Production publication still requires provenance, accuracy, curriculum, age suitability, accessibility, copyright, and editorial review.