Polygon (@0xPolygon), the network behind $POL, has switched on agent spend controls in its Open Money Stack (OMS). The feature lets a team set the limits within which an AI agent can spend, w
Polygon (@0xPolygon), the network behind $POL, has switched on agent spend controls in its Open Money Stack (OMS). The feature lets a team set the limits within which an AI agent can spend, with the agent's wallet enforcing those limits onchain.
A team approves what an AI agent can pay for, sets a $USDC budget and picks an expiry date. The agent's wallet contract then applies those terms. It rejects requests that go over budget, fall outside the approved list or arrive after the permission has expired.
The team can also revoke the permission at any time to stop further payments. If an agent's credit runs low, it can request a top-up without needing a fresh wallet approval each time.
The approach builds on features Polygon has already described for agent wallets. Polygon's Agent CLI announcement detailed session-scoped smart contract wallets with per-token spending limits, contract whitelists and 24-hour expiry. It also said the private key never enters the agent's context, which matters when a language model could be prompted to leak it.
OMS also supports a broader approval model. Polygon's overview of OMS says a user approves a permitted scope of actions once, and the app can then operate within that scope without asking for approval at every step.
Polygon has spent much of 2026 building payments infrastructure around stablecoins. Crypto.news reported that Polygon combined x402 with agent pay channels to process more than 11 million verified payment updates per second across a 25-hub test fleet, with final settlement anchored to Polygon Chain.
Polygon's developer documentation lists agentic payments as part of OMS, including x402 for pay-per-use APIs and ERC-8004 for onchain agent identity. The stated aim is infrastructure for autonomous agent commerce without human approval at every step.
Spend controls address the obvious risk in that model: an agent with open-ended access to funds. Capping the budget, restricting the list of approved payees and setting an expiry give teams a way to limit losses if an agent misbehaves or is manipulated. Because the rules sit in the wallet contract rather than in the agent's own logic, the agent cannot override them.
Polygon has also reported traction for scoped approvals. Bitzo, citing Polygon's blog, said roughly 8,000 wallets ran OMS's sign-on-once approval flow in the first half of 2026 across uses including yield, subscriptions, batch payouts and scoped AI agent spend.
Polygon: Polygon launches an onchain toolkit built for the agent economyPolygon Developer Docs: What is the Open Money Stack?Crypto.news: Polygon payment channels hit 11 million updates per second across 25 hubs
