Democratic Lawmakers Ask SEC How It Plans to Regulate AI Agents Trading on Brokerage Platforms

Democratic members of the House Financial Services Committee sent SEC Chair Paul Atkins a letter this week asking 13 questions about how the agency plans to oversee AI agents executing trades on registered brokerage platforms. The lawmakers, including Reps. Bill Foster and Brad Sherman (ranking members of the Financial Services and Capital Markets Subcommittees), cited risks of market volatility, correlated trading decisions, and what they called “herding behavior” among agents trained on similar data.

“The AI firms developing and deploying these agents have thus far operated largely outside the securities regulatory framework, even though their systems are making or enabling consequential investment decisions on behalf of retail investors,” the legislators wrote, according to Wealth Management.

The Regulatory Blind Spot

The letter lands as brokerage platforms split into two camps on agentic trading. Some are building frameworks to welcome third-party AI agents. Robinhood declared itself “open to agents” late last month, offering customers direct platform access for their AI agents and launching an “Agentic Credit Card” that lets agents spend on clients’ behalf. The firm has reported over 50,000 agentic trading accounts opened in the first weeks of launch.

Others are staying conservative. Joshua Broaded, head of AI and Advisory Services at compliance consulting firm ACA Group, told Wealth Management that the catalyst was OpenClaw: a free, open-source AI agent that users install locally. If someone connected an OpenClaw agent to their brokerage account, Broaded said, the platform likely never would have known.

“If you’re a brokerage firm and you’re signaling that you’re welcoming that, and you’re enabling trading strategies that leverage agents, it may be a way to pick up additional customers and enable something customers are interested in having,” Broaded told Wealth Management.

The Fiduciary Question

The lawmakers posed a pointed question to Atkins: will AI agents that facilitate trades be required to act in the best interest of the user, disclose conflicts of interest, maintain transaction records, safeguard customer information, or assume fiduciary duties? They also asked whether a third-party agent’s involvement would “alter, limit or absolve the broker/dealer of any obligations” under federal securities laws.

Traditionally, brokerage platforms argue that human customers make trading decisions, and the platform merely fulfills orders. AI agents complicate that framing. Broaded described it as an open question: platforms have “created the rails” and “invited the agents in,” but the customer is technically still the one deploying them.

The legislators gave Atkins until July 31 to respond. The SEC did not comment prior to Wealth Management’s publication.

Third-Party Vendor Risk

The letter also reflects growing concern about AI vendor security in financial services. Joshua Pantony, CEO of AI platform Boosted.ai, pointed to the Anthropic Mythos breach as a warning for the industry.

“The Mythos breach didn’t happen because Anthropic got hacked,” Pantony told Wealth Management. “It happened through a third-party vendor on the day of launch. Every financial firm running AI right now should be doing one thing: mapping exactly what data their AI vendors have access to and what the damage looks like if that vendor gets hit.”

Broaded noted that open-source AI options add another layer of complexity for regulators. Controlling frontier model providers like Anthropic, OpenAI, and Google DeepMind is one challenge. Preventing technically fluent users from circumventing any guardrails those providers build is another.

The Pressure Timeline

The letter arrives during a period of accelerating agent adoption in financial services. Taktile raised $110 million in Series C funding led by Goldman Sachs Alternatives for agentic financial decisioning. Palo Alto Networks’ Unit 42 recently identified novel financial threat classes in agent marketplaces, including agentic front-running and affiliate injection. Robinhood’s agentic trading accounts are scaling faster than its convertible note offering can finance.

Whether the SEC responds with new rulemaking, guidance, or silence will signal how quickly securities regulation adapts to a market where the trader is increasingly not human.