SEC Chair Atkins vows to deploy AI against criminals misusing AI

The SEC has announced plans to integrate AI technology into its operations in order to make detecting anomalies easier and faster and conduct risk assessments. 

The SEC is actively going after bad actors who exploit AI technology for fraudulent purposes and companies that engage in AI washing to deceive investors.

How is the SEC using AI to protect investors and catch fraudsters?

The Securities and Exchange Commission (SEC), under the leadership of Chairman Paul S. Atkins is implementing a strategy to “fight AI with AI.” The initiative is centered around the SEC’s AI Task Force, which was established to give the entire agency access to technological advances and ensure that the commission keeps pace with the rapid evolution of the private financial sector.

The Commission is using algorithms to detect market misconduct, including fraud and manipulative trading schemes. These tools can find anomalies in trading volume or price movements with greater speed and precision than traditional methods.

AI also helps the agency’s staff identify material omissions or misleading statements in documents filed by thousands of public companies more efficiently, allowing the SEC to react to public input and market changes in real-time.

Chairman Atkins has noted that the SEC’s objective remains to protect investors regardless of the tools it uses. This time, the agency is specifically looking out for signs of “AI washing.” This term is used to describe companies that make false, exaggerated, or misleading claims about their use of artificial intelligence to boost their stock price or attract investors.

What are the risks of using AI in government regulation?

One of the primary concerns regarding AI in government is the potential for black box decision-making, where an algorithm makes a choice without a clear, human-understandable reason. Chairman Atkins clarified that human interaction is necessary at every stage of the SEC’s risk assessment program.

“Due process demands it,” Atkins noted during a recent Financial Stability Oversight Council (FSOC) roundtable. An algorithm might identify a suspicious pattern or an anomaly, but it lacks the ability to determine the credibility of a witness or assess the intent of a market participant. Consequently, the final judgment remains with the Commissioners and professional staff.

Leading AI developers, including Google (Gemini), OpenAI, and Anthropic, have previously released reports detailing how malicious entities are exploiting their platforms. For example, OpenAI recently reported on disrupting state-sponsored threat actors who used AI to research vulnerabilities and generate phishing content. Similarly, Google’s Threat Analysis Group has tracked the use of Large Language Models (LLMs) in social engineering attacks designed to steal financial credentials.

The SEC will compel companies to disclose AI-related information if there is a substantial likelihood that a reasonable shareholder would find it important for an investment decision.

In early 2024, the Commission settled charges against two investment advisers for making false and misleading statements about their use of AI. In those cases, the firms claimed to use AI to analyze millions of data points to predict market moves, but the SEC found those claims to be false.

If you're reading this, you’re already ahead. Stay there with our newsletter.