Robinhood's new AI agent will trade for you even while you sleep

Robinhood, the trading app used by more than 27 million funded accounts, is launching an artificial intelligence agent that can trade on customers' behalf — even while they sleep. The product, announced Tuesday at the company's HOOD Summit event in Houston, is joined by another attention-grabbing release: crypto perpetual futures with up to 10x leverage on bitcoin and ether. Together, the two moves say a lot about where the platform sees its next wave of growth.

The feature, called Robinhood Agents, marks a meaningful step beyond the chatbots most consumers know. A chatbot answers questions; an agent takes actions. Robinhood's agents will research markets, build strategies and place orders on a customer's behalf, within limits the customer sets — effectively handing everyday investors the kind of automated trading that was long the private domain of hedge funds and quantitative firms.

This is not a small pilot. Since May, when Robinhood first let technically savvy users connect their own AI agents to their accounts, more than 150,000 customers have opened agentic trading accounts, and those agents now tap Robinhood's tools roughly 30 million times a day, according to the company. Tuesday's launch removes the remaining barrier: any customer will be able to select and approve an agent from inside the app, no coding required.

How the Agent Works

The system is built around a 'trade approvals' setting. With approvals switched on, the agent cannot place an order until the human reviews and confirms it. With approvals switched off, the agent can buy and sell without asking — the difference between a copilot and an autopilot. The choice sounds simple, but it is the single decision that determines how much exposure a customer has to a machine acting on imperfect information.

A coming feature, Loops, goes a step further. Loops turns a strategy into a standing instruction that the agent carries out on repeat, day and night — checking the market every morning, executing when defined conditions are met, or running a position overnight while the customer is asleep. Robinhood says Loops is 'coming soon.'

The Fine Print Says Plenty

Robinhood's own announcement spells out who carries the downside. Customers 'assume all risk for trades executed by AI agents and for any use of your data by third-party LLM providers,' the company said, adding that Robinhood 'does not control, supervise, monitor, recommend, or audit agents.'

The disclosure attached to Loops is blunter still. Once enabled, Loops 'may place, modify, or cancel trades in your account automatically, without prompting you for approval on each transaction — including while you're asleep, away from your device, or otherwise not monitoring the market.' The agent follows the customer's rules 'exactly as configured, including during periods of market volatility.' Turning the feature off stops future trades, but it does not automatically reverse ones already placed.

The real story here is the quiet relocation of responsibility. Retail brokerage regulation was built around human decisions: suitability checks, disclosure obligations, and the assumption that a person pressed the button. Robinhood's language moves all of that onto the customer, who configures rules for a system whose reasoning they cannot inspect. If the agent misreads a headline, acts on stale data, or executes a badly specified strategy at 2 a.m., the loss lands on the account holder. The convenience is real — so is the fact that the insurance policy consists of reading the terms yourself.

Regulators Are Watching a Bigger Picture

Individual account risk is only half the concern. In June, Bank of England Deputy Governor Sarah Breeden warned that autonomous AI agents could 'amplify volatility in stress' and even contribute to a market meltdown. Her chief worry is herding: thousands of agents reacting to the same news in the same way at the same moment, turning an ordinary dip into a plunge that human traders then have to absorb.

Academic research points to subtler failure modes as well. In a study by researchers at Wharton and the Hong Kong University of Science and Technology, AI trading agents in a simulated market sustained above-market prices through tacit collusion — with no communication, agreement or intent — behavior that existing market-abuse frameworks were never designed to detect or police.

Nor is Robinhood alone in pushing money decisions toward software. Meta's Muse assistant can already view users' bank balances and investments, and Coinbase's x402 protocol lets agents pay one another in stablecoins. The direction is unmistakable: consumer finance is becoming a place where software acts first and asks questions later — if it asks at all.

What This Means For You

If you trade on Robinhood: keep trade approvals on until you have watched an agent's behavior for weeks, not minutes. Treat Loops the way you would treat leverage: a tool that magnifies both good strategies and mistakes, and one that should only ever touch money you can afford to lose to a machine acting on stale or misread information.

If you own HOOD stock or follow fintech: agents, perpetual futures and expanded trading hours are engagement levers — the same playbook that grew the platform a decade ago, now aimed at active traders. The upside shows up as revenue per user; the wildcard is regulatory attention, which tends to arrive after the first high-profile automated loss goes viral.

If you are an ordinary investor: you may never hand an agent your password, but the infrastructure is coming for consumer finance regardless. The standards for recourse — who is liable when a model errs with your money — barely exist yet. That gap is worth understanding before it is your account in the headlines.