Delphia Securities Enforcement Case: The AI Advantage the SEC Said It Didn’t Have

Delphia agreed to pay a $225,000 civil penalty in a securities enforcement case over allegedly false claims that it used artificial intelligence and machine learning with client data in its investment process. The promised advantage was specific: clients’ collective data would help identify investments before other investors spotted them. If you worked on the data, the model or the marketing, you may recognize the gap between a capability sold to clients and the system available to deliver it.

The Securities and Exchange Commission announced the settlement on March 18, 2024, identifying Delphia as a Toronto-based investment adviser. This page follows Delphia’s client-data claim and the evidence that could test such a promise. Our broader page on AI claims in investment fraud compares advertised investment technology across different businesses.

Client Data Was the Promised Investment Advantage

The Securities and Exchange Commission alleged that Delphia made misleading statements about its AI capabilities from 2019 through 2023 in SEC filings, a press release and its website. According to the agency’s account, the claim concerned AI and machine learning that incorporated client data into the firm’s investment process.

The Commission reproduced Delphia’s description of that supposed advantage. The company’s claim, as quoted in the announcement, was that it:

That promise connected the client data to a smarter system and then to an investment decision ahead of the market. For someone who knows the operation, each connection raises a concrete question: what information entered the model, what the model did with it and how its output reached the investment process. We would examine those connections with the person who understands how the system worked when the claim appeared.

The Securities and Exchange Commission called the practice AI washing: marketing an AI capability that the adviser did not have. Clients choosing an adviser were being sold an AI investment advantage the agency said did not exist. For an employee who knows the data flow, the question is whether client information ever reached the investment decisions described in the pitch.

The Commission Challenged Delphia’s AI Advertising

According to the Securities and Exchange Commission’s announcement, its order found Delphia’s statements false and misleading because the firm did not have the AI and machine learning capabilities it claimed. The agency also charged a violation of the investment adviser Marketing Rule, which prohibits advertisements containing an untrue statement of material fact. Our explanation of how SEC complaints and orders differ covers what each kind of enforcement document says.

In the Commission’s account, the company offered clients an information advantage through a capability it lacked. If you reviewed that claim before publication, the technical description and the response to your review can explain how the final pitch took shape. We see a firm ahead of the market only in its advertising.

Global Predictions Made a Different AI Pitch

The same announcement from the Securities and Exchange Commission described a separate settled action against Global Predictions Inc., a San Francisco-based investment adviser. The agency said that firm’s website and social media carried misleading AI claims in 2023, including a claim to be the first regulated AI financial adviser and a claim about expert AI-driven forecasts. Global Predictions agreed to pay a $175,000 civil penalty.

The Commission also alleged that Global Predictions falsely claimed to offer tax-loss harvesting services and improperly limited liability in its advisory contract. The announcement’s $400,000 combined penalty covered both advisers. Delphia’s client-data representation and Global Predictions’ forecast and service claims give insiders different statements to compare with the actual product. A person who knows a forecast’s inputs may see a different gap from someone who knows which services clients received.

What Evidence Is Needed for Whistleblowing?

Useful evidence connects the dated AI claim to the data, model and investment process behind it. The securities fraud evidence trail often runs across public communications and internal systems. We work with the insider to understand how the records relate to the same claim.

In Delphia’s case, the Securities and Exchange Commission identified filings, a press release and the website as places where the alleged misstatements appeared. For a similar client-data claim, we would compare the wording and date with the data entering the system, the model version in use and the reviews made at the time.

The investment process supplies another comparison. We would examine how the claimed model’s output connected to investment decisions, using the relevant algorithms, code versions, transaction records and communications. A product or investment-operations employee can explain the handoff between an analytical result and an actual decision about a client’s money.

Review and approval records can explain how the public account developed. We would look at what the technical review said, what management understood and how the final description addressed any limitations. If you raised a concern about what the system could do, your knowledge of the response can connect the product’s limitations to a decision about what clients were told.

The Person Who Understands the Inputs

An insider’s contribution can be the explanation that makes a technical record intelligible. Someone who works with the data may know which inputs a model used, while someone in marketing or compliance may know how the resulting capability was described. We bring those perspectives together around the particular representation investors saw.

The 2019 paper Accounting Fraud: A Literature Review, published in Risk Governance and Control: Financial Markets & Institutions, examines statistical detection methods alongside whistleblowing and other methods drawing on people inside a business. That distinction is useful when we examine an AI investment claim: the person who understands the inputs and limitations can help explain what a model’s output actually means.

Whistleblower Rewards for Investment Fraud

The SEC whistleblower program can pay eligible whistleblowers 10% to 30% of qualifying sanctions collected where an action exceeds $1 million in monetary sanctions. We assess how your information fits the program with you. Where government money was cheated, a whistleblower case under the False Claims Act can pay 15% to 30% of what the government recovers.

When You Know What the AI Could Deliver

If you know how an investment adviser’s AI claim differed from the system your team worked with, tell our whistleblower experts what the technology could actually deliver. We develop your case around the data, investment process and decisions behind the claim, and bring in securities fraud whistleblower lawyers who understand those connections. Speak with us about building your strongest securities whistleblower case, with experts who stay beside you throughout it.