The use of algorithms in trading has been a decades-long practice, but recent advancements in artificial intelligence (AI) are presenting new challenges for regulators. While basic algorithms follow programmed commands, newer AI bots can learn from experience, synthesize vast amounts of information, and act autonomously when making trades.
The Evolving Landscape of Market Manipulation
Experts warn that one potential risk scenario involves collaboration between AI bots. Imagine hundreds of AI-driven social media profiles promoting narratives about specific companies. The information spread may not be fake but could be the amplification of existing news, influencing real social media users and potentially tipping the market. An investor’s robo-advisor could profit from this orchestrated narrative, even if they’re unaware of the scheme. This raises concerns about the effectiveness of current market manipulation regulations.
Current Challenges and Observations
Alessio Azzutti, assistant professor in law and technology at the University of Glasgow, notes that while this scenario is still hypothetical, less sophisticated schemes are already occurring, particularly in crypto asset markets and decentralized finance. “Malicious actors can be very active on social media platforms… encouraging members to invest in DeFi or a given crypto asset,” Azzutti explained. The spread of misinformation, even by unsophisticated actors, can ‘pollute chats’ and mislead retail investors.
The rapid, uncoordinated spread of market information online is also changing trading dynamics. Retail investors are more likely to follow trends rather than conducting their own analysis, which can destabilize the market and be exploited by AI bots. The GameStop saga is cited as an example of ‘herd trading,’ where Reddit users collectively drove up the stock price, catching big hedge funds off guard.
Regulatory Concerns and Potential Solutions
A spokesperson from the European Securities and Markets Authority (ESMA) acknowledged that the potential for AI bots to manipulate markets is “a realistic concern.” They highlighted the role of social media in rapidly spreading false or misleading narratives that can influence market dynamics. ESMA is actively monitoring AI developments.
Regulators face challenges in tracing collaboration between AI agents, as they don’t communicate in traditional ways. Itay Goldstein, professor of finance at the Wharton School, emphasized that regulation needs to adapt to these new strategies. Filippo Annunziata, professor of financial markets legislation at Bocconi University, suggested that supervisors need more sophisticated tools to identify market manipulation and proposed that AI developers include ‘circuit breakers’ in their tools.
The Path Forward
Experts discuss potential solutions, including designing AI to be more transparent and creating new laws around liability for AI deployment. The current legal framework faces challenges in assigning responsibility when AI acts maliciously without human intent, particularly in ‘black box trading.’ As Annunziata put it, “Supervisors tend to be tortoises, but manipulators that use algorithms are hares, and it’s difficult to catch up with them.”
