The Pulse
CFTC Chair Selig Raises ‘Know-Your-Agent’ Question for AI Trading
CFTC Chair Michael S. Selig raised questions about responsibility and oversight as autonomous software begins to act in financial markets. He said SI systems could help exchanges identify suspicious activity and that CFTC staff are using SI
AI.info Team · Written with AI · fact-checked ·
“Do we need a new flavor of know-your-customer, maybe KYA or know-your-agent?”
Michael S. Selig, chairman, Commodity Futures Trading Commission
Commodity Futures Trading Commission Chair Michael S. Selig put a regulatory question to the agency’s inaugural Frontier Forum in Washington on October 28: how should rules built around people, firms and financial intermediaries apply when software can identify trades and execute them for a user? Selig asked whether an agent’s developer might need registration, who would be liable if an agent “goes rogue,” and how regulators could identify the party behind an agent or group of agents.
The remarks raised possible issues for regulators; they did not propose a “know-your-agent” requirement or announce a change to CFTC rules. Selig also said his views were his own as chairman and did not necessarily reflect those of the Commission.
Rules built for identifiable market participants
The CFTC’s framework, Selig said, was developed around identifiable participants, intermediaries and institutions performing defined financial functions. Autonomous software complicates that model when it acts on a customer’s behalf, particularly if multiple agents or service providers contribute to a transaction.
“Our approach must remain grounded in the functions performed, the control exercised, and the economic realities of the transaction,” Selig said. That framing points to questions beyond the software’s identity: who directs it, what role its developer plays, and how much control the user retains.
Selig compared the current debate with earlier uncertainty in crypto markets, where software developers and entrepreneurs questioned whether publishing code or building user interfaces could trigger requirements designed for traditional intermediaries. He said the CFTC had worked to provide greater clarity in those cases, but did not spell out how that experience would translate into rules for trading agents.
AI could widen access—and link trading behavior
Selig also described potential benefits in derivatives markets, where firms use contracts to manage exposures such as crop prices, fuel costs and raw materials. AI systems could process information across thousands of instruments faster than a person and give smaller participants analytical tools that once required the resources of large institutions, he said.
The same capabilities bring risks. Selig warned that agents may react unexpectedly to market conditions, that systems built on similar models could produce correlated trading, and that faster autonomous transactions could make surveillance and risk management harder. He said SI systems can help exchanges identify suspicious activity, enable clearinghouses to monitor exposures more effectively, and improve regulators’ ability to detect misconduct. CFTC staff, he added, are already using SI to police markets.
A past trading proposal offers a limited precedent
The CFTC previously considered special controls for algorithmic trading firms after market disruptions including the May 2010 Flash Crash and Knight Capital’s runaway algorithm in August 2012. The agency proposed Regulation Automated Trading, but never finalized it; Selig said the proposal would not fit current technology well.
That history gives the agency a precedent for examining automated systems, not a ready-made rule for autonomous agents. Selig’s speech leaves open where responsibility should fall when software executes a trade: with the user, the developer, an intermediary, or more than one party.
For now, “know-your-agent” is a question the CFTC chair has raised, not a compliance standard the agency has adopted. Selig’s stated test is how a system functions, who controls it and the economic reality of its transactions.