About the author:

Outset Legal Lens is led by Alice Frei, Outset PR’s head of security & compliance. In this series, she draws on years of experience in legal, compliance, and due diligence work across Web3 projects to show where teams most often get it wrong, and how to build communication that supports growth without quietly creating liabilities.

‍

Earlier this year, the U.S. Securities and Exchange Commission (SEC) filed a complaint against Nathan Fuller, alleging that he raised approximately $12.3 million from around 150 participants through crypto-related investment programs. According to the SEC, he promised returns exceeding 100% in just 21 days while capping potential losses at 3%.

The regulator further contends that only about $380,000 – roughly 3% of the funds raised – was actually used to purchase crypto assets. Not to mention that the advertised AI-powered trading bots and risk management mechanisms didn’t exist as presented.

Whether the SEC proves allegations in court remains to be seen. Yet the case raises a question that extends far beyond a single enforcement action: Do such automated systems introduce a genuinely new category of investment products, or is AI simply becoming the latest justification for claims of predictable returns and limited risk?

‍

AI trading bots explained: What they do, and what they don't

The rise of artificial intelligence has transformed the way investors think about financial technology. Mention a trading bot, and most people immediately imagine a tool that can analyze vast amounts of data, identify opportunities faster than humans, and generate profits around the clock. 

In theory, the pitch sounds compelling: a machine learning system never sleeps, never acts on emotion, and can process information at a scale impossible for any individual trader. That perception, however, often blurs the line between what these bots really do and what investors believe they can do.

Legitimate algorithmic trading products have already entered financial markets. Depending on their design, they may assist with market analysis, pattern recognition, signal generation, portfolio optimization and protection, routine task automation, and trade execution. What they can't do is eliminate the fundamental realities of capital allocation. 

No model can remove market uncertainty, no algorithm can confidently predict future price movements, and no platform can consistently deliver positive returns regardless of prevailing conditions.

‍

Automated trading systems are not inherently problematic. The concern arises when businesses use the technology to suggest that downside risk can be engineered away.

‍

Why the Nathan Fuller сase looks surprisingly familiar

At first glance, AI-powered crypto trading forms a brand-new category of financial products. The tools may appear innovative, the terminology is different, and the marketing sounds more sophisticated than anything investors encountered in previous market cycles.

In fact, every cycle tends to develop its own narrative around projects with unusually high returns. Take crypto mining operations, arbitrage, lending programs, staking yields, secret market strategies, and proprietary algorithms, just to name a few. Today, machine learning models have topped that list.

That’s why regulators typically focus less on the underlying mechanics themselves and more on the overall arrangement. When evaluating AI-related products, they ask the same questions they have been asking for years:

• What exactly did the company/individual offer to participants?
• Who controlled the funds?
• Did the advertised strategy ever exist?
• Was an accurate risk disclosure in place?
• Were investor assets used as promised?
• Did the offering require registration under applicable securities laws?

In the civil action against Fuller, the SEC asserts that he engaged in fraud and conducted an unregistered securities sale. The central issue was whether he gave investors truthful information about how their money would be managed, how profit would be generated, and what risks they were taking.

‍

When does an AI trading product turn into a compliance trigger?

For regulators, there are certain types of communications that require heightened legal and compliance review:

• Guaranteed or predictable returns
• Fixed percentages of expected gains
• Claims that potential losses are limited or capped
• Promises of capital protection
• Statements that a system always outperforms the market
• Advertised win rates without independently verifiable evidence
• Backtested or simulated results disguised as real-world performance
• Licensing, insurance, or regulatory approvals that can’t be substantiated
• Assurances that AI removes human error or materially reduces uncertainty

The reason is straightforward: the representations like these shape investor expectations. In Fuller’s case, promotional materials included assertions such as "profitable returns are limitless" and "your risk is limited."

‍

The black box problem: How much should companies disclose?

Unlike traditional strategies, where investors may at least understand the basic logic behind a decision, AI products offer little transparency. Users rarely have an opportunity to inspect what’s under the hood, explore training methods, or check data sources.

This lack of visibility doesn’t automatically create a compliance problem. Nobody forces companies to publish proprietary source code, trade secrets, or the inner workings of every model they use. However, depending on the nature of the offering and the applicable legal framework, businesses may need to clearly disclose:

• Whether reported efficiency comes from live trading, simulations, or backtesting;
• The methodology supporting the calculation of performance results;
• Fees, assumptions, and other factors that may affect outcomes;
• Known limitations and risk factors;
• The extent of human involvement in decision-making;
• Historical drawdowns and adverse market scenarios;
• Custody arrangements and withdrawal procedures;
• Potential conflicts of interest and the handling of customer assets.

The less a user can independently verify, the more important accurate disclosure becomes.

‍

Where AI trading goes from here

One of the reasons AI trading bots are attracting increased legal attention is that regulators don’t need to develop an entirely new framework to evaluate their marketing claims. Rules governing securities fraud, advertising, advisers, consumer protection, and unfair commercial practices are already there. 

In separate cases involving Delphia and Global Predictions, the SEC accused investment advisers of making false and misleading statements about their purported use of AI. The message was clear: such communications can no longer remain unsubstantiated.

As a result, future scrutiny is likely to focus on issues such as AI washing, unsupported performance representations, synthetic or backtested results lacking appropriate context, undisclosed human involvement in supposedly autonomous systems, custody arrangements, and deceptive disclosures about potential downsides.

For companies operating in this space, compliance obligations cover not just official marketing documentation and channels – such as websites, pitch decks, and social media posts – but also founder accounts, Telegram and Discord communities, interviews, AMAs, and sales communications if they rely on claims about performance, potential losses, or what a technology-driven offering can deliver.

But none of this means that algorithmic trading has no long-term value. The challenge is separating genuine capabilities from narratives that present vague financial outcomes as fully controlled products.

The winners are unlikely to be the companies with the boldest AI manifestos. They will be those who can demonstrate what their systems do in practice, explain their limitations, be honest about their vulnerabilities, and show what happens when the model gets it wrong.

‍

This article is part of Outset Legal Lens. In this series, we’ll keep unpacking the legal side of Web3 communication, with a focus on helping teams speak clearly, responsibly, and in a way that supports the long-term growth of the industry.