Nebula Sentient Artificial Intelligence (SAI) Policy
Last updated: November 10, 2025
Purpose. This policy defines how Nebula SAI operates across BTCX products. “Sentient” is a brand term for high‑fidelity, multimodal, context‑aware behavior; it is not a claim of consciousness or personhood.
Scope & Architecture
- SI (Synthetic Intelligence). Real‑time trading helpers, arbitrage scanning, and yield optimization under hard risk budgets.
- AI (Artificial Intelligence). Risk management, predictive modeling, and embodied avatars (BTCX VXR).
- Control plane. AI proposes → human and/or policy guardrails approve → SI executes (or simulates).
Safety & Constraints
- No custody by default. No transfers without user‑signed transactions or revocable allowances.
- Risk budgets. Per‑trade and daily loss caps; instrument whitelists; circuit breakers.
- Human‑in‑the‑loop. Autonomy is opt‑in and visibly bounded by user‑defined budgets; global kill‑switch available.
- No guarantees / no advice. Analytics only. Users remain responsible for their decisions.
Data & Privacy
- Data minimization; encryption in transit and at rest; PII kept separate from telemetry.
- Retention windows with aggregation for research; audit logs are tamper‑evident.
Models & Updates
- Staged rollout (sandbox → canary → production) with measurable deltas and drift monitoring.
- Reproducibility: datasets, model hashes, and prompts are versioned; changes are logged.
- Red‑team testing for manipulation, leakage, and injection.
Blockchain Anchors
- Bitcoin: policy hash published as a timestamped “relic”.
- Ethereum: governance registry emits
PolicyUpdated(root, uri, version) (EIP‑712 signatures).
- Solana: high‑frequency heartbeats for near‑instant global synchronization.
Incident Response
- Automatic pause on abnormal loss, stale feeds, or signature mismatch.
- On‑chain kill‑switch; public post‑mortem and rollback to last good root.
Disclaimer: Informational only; not legal or investment advice.
}