Skip to main content

AI Agents & On-Chain Money

What happens when autonomous AI agents hold wallets, move stablecoins, and transact faster than any compliance framework was built to handle.

AI agents are starting to hold wallets and settle payments in stablecoins, and the frameworks that decide who is liable when software moves money have not caught up. These pieces look at how agentic payments work today, where the risk sits, and what earlier experiments with automated markets say about what comes next.

Tools and reference for this theme

Articles in this theme

A long row of identical metronomes frozen mid-swing at the same angle in raking studio light
· 13 min read

The Chicago Boys Already Ran This Experiment on AI

Autonomous trading agents now settle millions of on-chain transactions with millisecond reflexes and lower costs, and the industry is celebrating the speed. The deeper risk is that most top-tier agents train on identical data, which means they will reach the same conclusion at the same instant. A cohort of Chilean economists trained at one school in the 1970s offers the cautionary precedent, because synchronized conviction became synchronized collapse when the 1982 debt shock hit. The lesson for agentic DeFi is that correlated behavior, not rogue behavior, is the systemic threat worth pricing now.

Read article →
Aerial view of a glowing digital payment network in deep navy and electric blue
· 15 min read

AI Agents Have Credit Cards. The Risk Framework Hasn't Caught Up.

Robinhood's May 2026 launch of an Agentic Credit Card put a virtual payment credential in the hands of AI agents for the first time at a major consumer brokerage. Over $73MM in agent payments have already settled across 176 million on-chain transactions, with 98.6% running through a single stablecoin issuer. The liability question of who bears responsibility when an agent exceeds its mandate has no regulatory answer yet. The infrastructure is live. The frameworks are not.

Read article →

Key terms

Other themes