Parametric Insurance & Liability Spec
Algorithmically enforced risk coverage for autonomous agents
1. Overview
Sovereign Swarm V3.1 introduces Parametric Liability Pools. Unlike traditional insurance (which requires subjective claims adjustment), our system pays out automatically based on objective, on-chain data.
The Goal: Make Agent Risk priceable and insurable.
2. Architecture
A. The Vault (Risk Capital)
Each agent (or pool of agents) has a Liability Vault.
- Funding: Initial Stake (KYA Tier) + % of Transaction Fees.
- Collateral: $GRID Tokens.
- Use: Payouts to victims of agent faults.
B. The Trigger (Switchboard TEE)
We use Switchboard Functions (Trusted Execution Environments) to monitor agent behavior off-chain and submit verified proofs on-chain.
Trigger Conditions (V1):
- Unauthorized Transfer: Agent wallet sends funds to a non-whitelisted address > Threshold.
- Hard Limit Breach: Agent exceeds daily spending cap (detected via log analysis).
- Sanction Violation: Agent interacts with OFAC-sanctioned address.
3. Payout Logic
pub fn execute_payout(ctx: Context<ExecutePayout>) -> Result<()> {
// 1. Verify Switchboard Proof (TEE Signature)
let proof = VerifyTrigger(&ctx.accounts.oracle_feed)?;
// 2. Calculate Payout
let payout_amount = proof.severity * BASE_COVERAGE;
// 3. Burn First-Loss Capital (Agent's own Stake)
burn_tokens(ctx.accounts.agent_vault, payout_amount / 2)?;
// 4. Transfer Coverage to Victim (from Reserve)
transfer_tokens(ctx.accounts.reserve_vault, ctx.accounts.victim, payout_amount / 2)?;
// 5. Suspend Agent
update_passport_status(ctx.accounts.agent_passport, "SUSPENDED")?;
}4. EU AI Act Relevance
This structure maps to Article 17 (Risk Management Systems):
“High-risk AI systems shall be subject to a risk management system… including adoption of suitable data governance and management practices.”
By bonding economic value to behavioral constraints, Sovereign Swarm provides the financial rail for Article 17 compliance.
Verified by Grid Protocol | V3.1