Market manipulation is an old issue. People try to make money off unsuspecting investors by artificially influencing the stock price. But what if the person manipulating the markets is not a human?
Mary Louise Kelly, host:
As long as financial markets have existed, market manipulation has also existed. But what if the people engaging in market manipulation aren’t people at all? Indicators, Adrian Ma and Willen Wong reveal how AI could disrupt financial markets.
ADRIAN MA, BYLINE: There are a lot of ways AI can be used to manipulate financial markets, but to help us understand them, we’ll categorize them into two basic groups. The first is what you might think of as human-driven manipulation.
WAYLEEN WONG, BYLINE: And to help us explain, we called Nicole Turner Lee. She is director of the Center for Technology Innovation at the Brookings Institution. In her research, Nicole demonstrated the ways in which artificial intelligence can be used to tamper with markets. A big one of them is spreading misinformation.
Nicole Turner-Lee: We see this every day in the analog world – the president makes announcements about tariffs, which creates market disruption. When you have something embedded in a technical system, the scary thing about it is that you may not know where it came from.
Wong: Generative AI makes it much easier for bad actors to manufacture disinformation. With just a few keystrokes, anyone can create a fake news article or deepfake recording. With the help of bots, they can easily spread misinformation online.
MA: But what if AI could use AI to manipulate markets? What? What we’re talking about here is that AI-powered trading bots run amok. Now, of course, trading bots are nothing entirely new. For many years, hedge funds have been using them to execute high-frequency trades. But as finance professor Ekaterina Svetlova says, these robots still require a lot of human input to tell them how to trade.
Ekaterina Svetlova: In the case of high-frequency trading, the machine was given clear rules of what to do. Now we have algorithms that do not receive clear rules from humans.
WONG: At the University of Twente in the Netherlands, Ekaterina studies how technology is shaping financial markets, and says these new algorithms have given rise to a smarter type of trading bots powered by AI machine learning. In particular, something called reinforcement learning. This is where the AI agent is given a goal, such as maximizing profits in the long run, and without any additional instructions from a human, the AI goes to work.
MA: Researchers at the University of Pennsylvania ran a simulation to see what would happen if they unleashed a group of AI robot traders powered by reinforcement learning on the market. What they found was that instead of trading against each other, as you would expect in a competitive market, these bots began colluding with each other to manipulate the market. The strange thing about this is that if humans engaged in this kind of collusion and manipulation of financial markets, they would be breaking the law. But the crime of collusion and market manipulation historically required human intent.
Wong: So who’s responsible if a gang of trading bots engages in a financial crime spree? This is a legal gray area, and regulations may be needed to fix that, says Nicole Turner-Lee of the Brookings Institution. Like all the experts we spoke to, Nicoll says AI is not an inherently bad thing for financial markets. It can be used, for example, as a fraud detection tool.
MA: But as regulators continue to catch up, that means a lot of power rests with financial companies that are piloting these AI systems and still figuring out what they can do.
Wong: Eileen Wong.
MA: Adrian Ma, NPR News.
Copyright © 2025 NPR. All rights reserved. Visit our Terms of Use and Permissions pages at www.npr.org for more information.
Accuracy and availability of NPR transcripts may vary. The transcript may be revised to correct errors or match updates to the audio. Audio on npr.org may be edited after its original broadcast or publication. The official record of NPR programming is the audio recording.