CFTC chair Selig warns AI agents could trade crypto, launch tokens, raise regulatory issues

Commodity Futures Trading Commission Chairman Michael Selig stated that autonomous artificial intelligence agents may soon be capable of managing finances, trading digital assets, and even creating their own crypto tokens. Selig delivered these remarks at the agency’s inaugural Frontier Forum held in Washington, DC.

AI in Everyday Finance

Highlighting current advancements, Selig remarked that technologies once considered science fiction, such as refrigerators independently purchasing groceries using funds earned from staking Ether or creating digital tokens, are now within reach.

“It’s no longer science fiction to imagine your refrigerator autonomously purchasing your groceries with funds that it earned from staking ether or launching a meme coin,” Selig said.

He emphasized that these views reflect his personal perspective and not necessarily that of the full Commission. Selig used the term “super intelligence,” noting it as an alternative to “artificial intelligence,” and referenced President Trump’s preference for this terminology.

Selig outlined the rapid adoption of AI-powered tools across the finance sector. He cited apps like Claude, ChatGPT, and Gemini, which now assist users in analyzing financial data and can interface with various software systems. Brokerages have begun integrating these technologies for their clients, and some individuals already allow AI agents to access and manage their banking or brokerage accounts.

Selig also pointed out that blockchain technology enables autonomous agents to hold, trade, and transfer cryptocurrencies without reliance on traditional financial institutions, further reducing the need for intermediaries.

Investor interest in AI-powered financial tools continues to surge. According to an eToro survey published in September, 79% of US millennial investors and 77% of Gen Z investors have used or would consider using AI for investment decisions.

Regulatory Questions and Risk Considerations

Selig raised several regulatory concerns stemming from the emergence of autonomous agents. He explained that current CFTC rules presuppose identifiable human market participants, whereas autonomous software challenges that framework.

“Does the developer of that agent become subject to registration? Who is liable if the agent goes rogue?” Selig proposed. He also questioned whether regulators should introduce “KYA” or “know-your-agent” protocols for such autonomous trading systems.

He cautioned that these agents could behave unpredictably during volatile markets and that similar AI models might cause herding effects, complicating market surveillance. The Bank of England’s Financial Policy Committee issued a similar warning last year, suggesting that the widespread use of comparable models could amplify market shocks.

Selig suggested that AI could give smaller traders and businesses access to risk management tools previously available only to large institutions. He confirmed that CFTC staff members are already deploying super intelligence technologies to monitor trading activity.

Animoca Brands Chairman Yat Siu also noted that if a large number of people query the same AI about a trading decision, the agents might generate similar results, raising further concerns about systemic risk.

Mini dictionary: Animoca Brands is a Hong Kong-based company known for its investments and development in blockchain gaming and digital property rights, especially in Web3 and non-fungible token (NFT) projects.

Industry Response and Future Proposals

Selig highlighted that cryptocurrency exchanges are already building agent-driven tools for users. For instance, Coinbase introduced Agentic Wallets in February, allowing users to authorize autonomous agents to make on-chain transactions within pre-set limits. In March, Binance launched AI Agent Skills, which enable users to access real-time market data and execute spot trades via AI-driven agents.

Reflecting on regulatory history, Selig referenced the CFTC’s earlier Regulation Automated Trading proposal, which followed the 2010 Flash Crash. However, he noted that the rule was never finalized and believes it would not suit today’s market landscape.

The CFTC recently invited public comments on the development of compute derivatives, potential financial instruments that could help firms hedge their computing costs. Selig reported receiving over 100 submissions, indicating significant industry interest and confirmed the agency’s intention to evaluate the feedback seriously.