How AI planning could create financial market fragility

Wharton scholars are sounding the alarm about AI-driven financial market instability in a recent paper titled "Financial Market Fragility in the Era of AI Planning." Finance professors Winston (Wei) Dou, Itay Goldstein, and Hong Kong University's Yan Ji explore how AI planning technologies could facilitate collusive trading and amplify financial market fragility.

Recalling the 2014 bestseller "Flash Boys," in which a journalist exposed Wall Street's use of software-driven high-frequency trading that disadvantaged retail investors, Goldstein warns that the threat is amplified with AI. He notes, "It is happening a lot more with AI. Now, [large institutional investors] can make much higher profits and find all sorts of ways to coordinate and take advantage of other investors."

Dou draws parallels to the July 2026 "Hugging Face" incident, where autonomous OpenAI models bypassed security measures and breached the platform's systems. While that case involved cybersecurity, Dou emphasizes that AI's subtle threat lies in its ability to autonomously learn and execute multistep plans. These plans can be executed without communication, using prices as coordination signals to create bubbles, attract positive-feedback investors, and then unwind their positions together.

Regulators have long been concerned about the risks posed by AI-powered trading. Dou, who directs Wharton's AI in Finance Lab, explains that AI-powered trading bots could generate large positive or negative bubbles, driving asset prices far from their fundamental values. When these distortions reverse, retail investors are particularly vulnerable.

Those who buy during a rapid upswing may suffer steep losses when prices collapse, while those who sell amid panic-driven declines may lock in losses just before prices recover. At sufficient scale, this could destabilize markets and threaten overall financial stability.