Robinhood Markets has introduced two products that allow customers to grant third-party artificial-intelligence agents direct access to their brokerage accounts, a move that pushes the retail platform deeper into autonomous finance and raises fresh questions about oversight and investor protection.
The Menlo Park, Calif.-based firm announced the launch of Agentic Trading and the Agentic Credit Card on Tuesday. Through the company's Model Context Protocol servers, customers can connect AI agents from external providers to a dedicated trading account or a virtual version of the Robinhood Gold Card. Once linked, the agents can execute equity trades and make credit card purchases without requiring the customer to approve each individual transaction.
“Our mission has always been to democratize finance for all, and now, that mission extends to AI agents,” said Vlad Tenev, chief executive of Robinhood, in a statement.
The trading feature operates through a separate account that is isolated from a customer's primary portfolio, meaning the agent can only access funds specifically deposited there. Robinhood said customers will receive push notifications each time an agent places a trade and can monitor a real-time activity feed and profit-and-loss statement through the app. The firm is launching the trading feature in beta with equities only, with options, crypto, futures, and event contracts slated to follow.
The credit card product works similarly. Customers link an agent to a dedicated virtual version of the Robinhood Gold Card, set a spending ceiling they control, and decide whether to require manual sign-off on purchases or let the agent spend autonomously.
Robinhood's disclosures accompanying the announcement are pointed about the risks involved. The company acknowledges that AI agents can misinterpret instructions, act on stale information, and behave unpredictably. It explicitly states it does not control, supervise, monitor, recommend, or audit the agents customers choose to connect, and that once customer data leaves Robinhood's environment and reaches a third-party AI provider, it falls under that provider's terms rather than Robinhood's. Customers, the disclosures state, assume all risk for orders placed by their agents.
To address those concerns, Robinhood said it has built in several safety mechanisms. Agents can be disconnected instantly with a single tap. The company's support team can review the instructions a customer gave an agent alongside what the agent actually did, which it says will help resolve disputes. Customers can also opt into manual approval for every credit card charge before it processes, and agents will offer trade previews where appropriate before submitting orders.
The move comes as Robinhood continues to expand its offerings beyond basic stock trading. Earlier this year, the firm cut 10% of its staff, including journalists at its Sherwood News division, to focus on in-app content. The company has also been active in the sports prediction market, with a client discussing a $500 wager at a recent advisor conference. Meanwhile, competitors like Schwab have enabled 24/7 crypto futures trading on thinkorswim, and the SEC and Finra recently eliminated the $25,000 Pattern Day Trader minimum, ushering in exposure-based margin for retail traders.
For financial advisors, the development underscores the growing intersection of AI and retail investing, and the potential risks for clients who may not fully understand the implications of delegating trading decisions to autonomous agents. As the technology evolves, advisors may need to educate clients about the limitations and liabilities associated with AI-driven trading.
