CFTC chief questions how to police AI trading but calls it a path toward autonomous markets
Regulator weighs AI in trading at CFTC forum
The U.S. Commodity Futures Trading Commission held a forum on Wednesday in which Chairman Mike Selig raised questions about how to regulate artificial intelligence in financial markets.
Selig said AI, crypto, and modern financial infrastructure could change markets by expanding access to complex financial tools, helping businesses manage risk more precisely, and creating new kinds of economic activity.
At the same time, he said regulators still need to answer hard questions about how to oversee computer agents that trade on behalf of users.
Questions regulators still need to answer
- If an AI agent spots trading opportunities and executes trades for a person, should the developer of that agent be required to register with the regulator?
- Who is responsible if such an agent starts acting on its own?
- How can regulators find out who a non-human trading entity is really working for?
- Should regulators create a new identification rule, something like "know-your-agent," instead of the usual "know-your-customer" checks?
What the CFTC chairman said
Selig described the technology in strong terms, comparing it to major past innovations such as the printing press, the steam engine, and the internet. He argued the label "artificial" is no longer fitting and joined other panelists in using "superintelligence," a term promoted by President Donald Trump.
He said the technology continues a decades-long trend toward trading systems that need less human control.
On the enforcement side, Selig noted the agency's staff already relies heavily on AI to monitor derivatives markets. Derivatives are financial contracts whose value comes from another asset, such as a commodity, currency, or stock index.
He also pointed to crypto as a possible guide. He said the CFTC has already dealt with similar uncertainty around whether writing code or building user interfaces should trigger rules written for traditional financial middlemen, and that he has pushed to make the answers clearer.
What remains open
The forum raised more questions than it settled. No new rules or timelines were announced, and Selig did not say whether developers of trading AI would be required to register.
Why this matters
As software agents become more common in trading, regulators will need new answers about accountability and identification. How they choose to act could shape who can build such tools, who can use them, and what safeguards will be required.