Skip to content
Trading Technology

Robinhood Is Handing AI Trading Agents To The Mass Market - With An Off Switch For Human Approval. The Bank Of England Has Already Said What Worries It

Robinhood is now rolling out in-app AI trading agents to customers in randomised waves, after more than 150,000 people opened agentic accounts for outside agents since May and those agents began calling Robinhood's tools around 30 million times a day. A customer names an agent, funds a separate account and connects a model from Anthropic or OpenAI; the agent analyses markets, builds strategies and trades. Trade-by-trade approval is on by default - and can be switched off. Alongside: crypto perpetuals at up to 10x, earnings contracts built on Cboe's KPI binaries, which face an SEC deadline on 13 October, and 24/7 weekend equities planned for early 2027. In June the Bank of England's Sarah Breeden warned that agents trained alike could react to a shock in lockstep, and that a human in the loop for every action is unlikely to be realistic. Here is what retail agentic trading changes for investors, brokers and the platforms that will have to support it.

AlchmAI Editorial12 min read

150,000+

Agentic accounts opened since Robinhood let outside AI agents trade in May; agents now call its tools about 30 million times a day

Default: on

Trade-by-trade approval for Robinhood Agents - which customers can turn off for fully autonomous trading

10x

Maximum leverage on Robinhood's coming crypto perpetual futures for Bitcoin and Ether, 3x on other coins

13 Oct

SEC deadline to approve, reject or extend review of Cboe's KPI binary contracts, which Robinhood will offer first

Retail AI trading has a mass-market product now. After announcing Robinhood Agents at its HOOD Summit in Houston on 30 September, Robinhood is rolling the feature out to customers in randomised waves over the coming weeks. Setting one up takes three steps: name the agent, open a dedicated agentic account, and connect a model - initially from Anthropic or OpenAI, with OpenAI's GPT-Luna free through the end of 2026. The agent can analyse markets, build strategies and trade on the customer's behalf. Robinhood is building on scale it already has: since it opened its servers to outside agents in May, more than 150,000 customers have opened agentic accounts and agents call its tools roughly 30 million times a day.

The design choices are deliberate. Agents can only reach money placed in the separate agentic account. Trade-by-trade approval is on by default - but customers can switch it off and let the agent trade autonomously. 'Loops', standing instructions executed around the clock, are coming, as are paid 'Agent Apps' data feeds from eleven partners. Robinhood is candid about expectations - a product executive said very few people will put a large portion of their net worth in an agentic account and start trading on day one - and about responsibility: customers are accountable for their agents' decisions, and the company has not yet compared outcomes between agentic and traditional trading.

The Rest Of The Launch

  • Crypto perpetual futures on Bitcoin, Ether, Solana, XRP and others, at up to 10x leverage on BTC and ETH and 3x on the rest, with 0.01% fees through 2026.
  • Earnings contracts built on Cboe's planned KPI binary options, which let traders take positions on specific company metrics. Cboe plans to list contracts on 23 US companies, subject to SEC approval; the regulator's deadline for a decision is 13 October.
  • 24/7 weekend trading in US stocks and ETFs, planned for early 2027 pending approval, alongside extended options hours, one-cancels-the-other orders and 4x intraday margin for eligible customers.

What It Means For Investors

  1. 01Keep approvals on until you have watched the agent through a bad week as well as a good one. The off switch exists; it does not have to be used.
  2. 02Fund the agentic account with money you can afford to lose. The separate-account design is a genuine protection - use it as one.
  3. 03Treat leverage and agents as a dangerous pairing. A 10x perpetual and an autonomous agent that reacts to the same headline as thousands of others is the scenario Breeden described.
  4. 04Understand what the agent reads. Paid data feeds are an edge and an exposure; an agent acting on a feed is only as good as the feed.

“Robinhood has made agentic trading easy to start. Whether it is safe depends on the settings customers choose and the limits platforms enforce - and the central bank has already said which setting worries it.”

What It Means For Brokers And Trading Platforms

Customers will expect agents from every broker within a year, and the platforms that build them well will be judged on the controls, not the chat. That means hard limits the agent cannot change - position size, daily loss, leverage, permitted instruments - enforced by the platform; approvals that revert to 'ask first' automatically when markets are disorderly; strategies tested before they run unattended; and, above all, visibility. A customer should be able to open a chart and see every proposal the agent made, every order the platform blocked and every fill, plotted against the price. Real-time charting of agent decisions is how trust is built and how complaints are resolved.

The Bottom Line

Robinhood is rolling out in-app AI trading agents to the mass market, building on 150,000 agentic accounts and 30 million daily agent tool calls, with approval on by default but switchable off, alongside 10x crypto perpetuals, Cboe-based earnings contracts awaiting an SEC decision on 13 October and weekend equities planned for 2027. The Bank of England's warning that agents trained alike could amplify stress in lockstep is the right lens. For investors, keep approvals on, fund cautiously and avoid pairing agents with leverage; for brokers and platforms, the product is the controls - enforced limits, stress-mode approvals and real-time charts of every agent decision. That is the trading platform and charting work we build in London, and retail agentic trading has just made it urgent.

References & Further Reading

Trading Platform DevelopmentTrading Workflow automationAgentic AIRobinhoodTradingView & Custom Chart LibrariesHigh-Frequency Charting SolutionsAgentic AI London
Share Email
AI

AlchmAI Editorial

Research and analysis, London

The AlchmAI team writes about the markets, technology and regulation we work with every day. We build trading platforms, real-time charts and AI analysis tools for brokers, prop firms and fintech teams from our office in Mayfair, London. Every article lists its sources. Nothing we publish is investment advice.

This article is general information and commentary. It is not investment advice or a recommendation to buy or sell any investment. Important information

Keep reading

All articles