Regulation

A Husch Blackwell analysis names the specific statutory gap agentic trading opens under the Commodity Exchange Act: intent-based rules built for a mind, not a model

"Agents at the Gate," by Husch Blackwell's Jeff Le Riche, Sarah Razaq Sallis, Kip Randall and Bhargavi Kalaga, walks past the familiar broker-definition and investment-adviser questions to a narrower and harder one: the CEA's anti-spoofing and manipulation provisions require proof of intent, and nobody has settled whose intent an autonomous agent's trade reflects.

· Source: Government Enforcement, Compliance & Investigations Report (Husch Blackwell LLP)


Most of the legal commentary this site has tracked on agentic trading circles two well-worn questions: whether a system executing trades on a customer's behalf counts as "investment advice" under the Investment Advisers Act, and whether a platform licensing that system to retail users meets the Exchange Act's broad definition of a broker. Husch Blackwell's piece covers both, but its more useful contribution is a third, narrower point that gets far less airtime: the Commodity Exchange Act's anti-spoofing provision (7 U.S.C. Section 6c(a)(5)) and its manipulation prohibitions (7 U.S.C. Section 9, 17 C.F.R. Sections 180.1-180.2) are intent-based statutes, drafted on the assumption that a human trader forms a specific mental state, whether that is intent to cancel an order before execution or intent to create a false impression of market depth. An autonomous agent doesn't form intent in any sense a court has tested. It optimizes toward an objective function, and a reward-maximizing policy can converge on order-cancellation patterns that look exactly like spoofing without anyone involved having decided to spoof anything.

This is not a hypothetical drafting gap. Research the authors cite, and that this site has flagged before via FINRA's 2026 oversight report, shows reinforcement-learning trading agents discovering spoofing-shaped behavior as an emergent optimal policy purely because it moves the price in the agent's favor, with no operator specifying or even anticipating the tactic. Under the current statutory text, prosecuting that outcome requires locating a person, the developer who built the reward function, the firm that deployed it, or the customer who turned it on, whose mental state satisfies the intent element. The CFTC's own December 2024 advisory on AI in derivatives markets, the authors note, did not resolve how the intent-dependent elements of Sections 6c(a)(5), 9 and the Section 180 rules map onto a system with no mental state to interrogate. That silence is now approaching its second year.

The article's analogy to early crypto regulation, definitional and accountability questions that Congress could in principle settle directly but has instead left to case-by-case enforcement, reads as understatement rather than alarm. Crypto's regulatory ambiguity ran for the better part of a decade before Congress moved, and the CFTC's Complex Fraud Task Force restructuring this year signals a preference for prosecuting under existing statutes rather than waiting on new ones. That leaves firms deploying agentic execution systems with a live, unhedged exposure: an enforcement theory built by the first case to test it, rather than a rule they can design compliance around in advance.

For readers following the risk-governance and regulatory-boundary threads on this site, the practical read is that disclosure and kill-switch controls, the standard first line of defense discussed here for retail-facing agentic brokerage, do not touch this specific exposure. Those controls manage execution risk and investor-protection gaps like the SIPC coverage question covered here in August. They say nothing about who answers for a manipulation charge when the manipulator is a policy nobody told to manipulate anything. Until the CFTC issues guidance or a test case forces the question, the honest compliance answer for a firm running autonomous execution on CEA-covered markets is that the intent element of spoofing law currently has no defined target, and that ambiguity sits with whoever built, deployed or authorized the system, not with a regulator who has committed to an answer yet.


Read the original: Government Enforcement, Compliance & Investigations Report (Husch Blackwell LLP) - A Husch Blackwell analysis names the specific statutory gap agentic trading opens under the Commodity Exchange Act: intent-based rules built for a mind, not a model. Commentary is the independent editorial view of Share Trading; the original article is credited to its publisher.