Skip to main content

What are the disclosure rules for AI-generated financial content?

Regulators are scrambling to keep pace with AI's spread into financial services, but the rules are still incomplete. As of 2024, there is no comprehensive federal mandate requiring every AI-generated financial analysis, newsletter, or recommendation to be labeled as such. The SEC has issued guidance on some AI uses. The FTC has warned about deceptive AI claims. But the gaps are large, and enforcement is sporadic. This creates a situation where an investor can read what appears to be a human-written financial analysis without realizing it was generated by a machine. Understanding the current regulatory landscape — and its limits — helps you navigate AI-generated financial content more skeptically.

Quick definition: AI disclosure rules are regulations requiring companies or individuals to reveal when artificial intelligence is used to create financial content, investment recommendations, or trading strategies; current rules are partial and inconsistent across jurisdictions.

Key takeaways​

  • The SEC has guidance on AI use in investment management but no blanket requirement to disclose when content is AI-generated.
  • The FTC prohibits deceptive marketing and has sued companies for making false claims about AI capabilities, but has not established clear financial AI labeling requirements.
  • Some states (like California) have enacted laws requiring disclosure of AI use in certain contexts, but these are inconsistently applied to financial content.
  • Most AI financial tools rely on disclaimers buried in terms of service rather than prominent disclosure.
  • International regulators (EU, UK) are moving faster than the U.S. toward requiring AI transparency in financial services.
  • The lack of clear rules creates an incentive for operators to avoid mentioning AI at all, leaving investors in the dark.

Current SEC guidance on AI​

The U.S. Securities and Exchange Commission has issued limited guidance on AI use in investment management. The main document is the "Risk Alert on Artificial Intelligence" (issued in 2023), which addresses how investment advisors and asset managers should use AI responsibly. Here is what the SEC currently requires or recommends:

For registered investment advisors (RIAs): The SEC expects that advisors using AI to make investment recommendations must have adequate governance. This means:

  • The firm should have written policies explaining how AI is used.
  • Someone at the firm should be responsible for overseeing AI quality and bias.
  • The AI system should be tested regularly to ensure it is not producing discriminatory recommendations.
  • The advisor must disclose material conflicts of interest — for example, if the advisor owns the AI company being used.

But the SEC does not require investment advisors to disclose to clients that they use AI. The SEC expects advisors to comply with their general fiduciary duty to disclose material facts. In the SEC's view, using AI is not inherently material — it only becomes material if the use creates a conflict of interest or reduces the quality of advice.

This creates a loophole: an investment advisor can use an AI system to generate recommendations, and as long as the recommendations are sound and there is no conflict of interest, the advisor does not have to tell clients "we used AI." This is problematic from a transparency standpoint.

For robo-advisors and fintech platforms: The SEC treats these as investment advisors, so the same rules apply. But robo-advisors often claim they are not "advisors" at all; they are merely software tools providing information. If a robo-advisor successfully argues that it is not a registered advisor, then SEC rules do not apply to it. This is a major gap. Many robo-advisors operate in this gray zone, providing AI recommendations without registering as advisors and without complying with fiduciary rules.

FTC enforcement and deceptive marketing​

The Federal Trade Commission has broader authority to regulate deceptive practices in commerce. It cannot require AI disclosure, but it can sue companies for making false claims about AI capabilities. A few recent actions illustrate the FTC's approach:

Example 1: False AI capability claims. An AI trading app advertised itself as using "advanced machine learning to achieve 20% annual returns." The FTC sued, arguing that the company had not actually backtested the system and could not prove its claims. The company settled and agreed to substantiate its performance claims with real data.

Example 2: Deceptive origin claims. A newsletter startup said its content was "written by an elite team of financial analysts," but the team was actually a single person using an AI tool. The FTC sued for misrepresenting the content's origin. The company settled.

Example 3: Conflict-of-interest concealment. An AI recommendation tool suggested users buy certain stocks, but failed to disclose that the tool's operator had ownership stakes in those stocks. The FTC investigated for undisclosed conflicts of interest.

The FTC's approach is reactive: it sues after finding deception. It has not proactively required AI disclosure across the financial industry. But the threat of FTC action does deter the most egregious deceptions.

State-level rules​

Some U.S. states have enacted laws requiring disclosure of AI use in certain contexts. The most prominent example is California's AB 701, enacted in 2023, which requires disclosure of "synthetic media" (deepfakes, AI-generated audio/video) in political advertising. However, this law does not apply to financial content; it is specific to political speech.

A few states are considering broader AI transparency laws. Colorado and New York have proposed legislation that would require AI disclosure in hiring and lending decisions, but these focus on non-financial uses and have not been broadly adopted across the country.

One area where state regulators have moved is insurance. Some states require insurance companies to disclose when they use AI to make coverage or pricing decisions. This is because insurance regulators saw AI creating systematic biases that harmed consumers. Similar logic could apply to financial advice, but few states have extended it that far.

The upshot: state rules are fragmented and do not yet comprehensively address AI-generated financial content.

International regulators moving faster​

The European Union and the United Kingdom are ahead of the U.S. in mandating AI transparency. The EU's AI Act (passed in 2023, implementation ongoing) requires transparency for high-risk AI systems, including those used in financial services. Financial institutions deploying AI must document:

  • What data the AI was trained on.
  • How the AI makes decisions (explainability).
  • What safeguards are in place against bias.
  • How the system is monitored for errors.

These requirements must be disclosed to regulators and, in some cases, to customers.

The UK's Financial Conduct Authority (FCA) has issued guidelines requiring firms using AI in financial services to explain its use to customers if the AI is material to the service. This is closer to a transparency requirement than anything the SEC has done.

If these international rules tighten and become more stringent, U.S.-based AI financial tools might face competitive pressure to disclose more, even in the absence of U.S. regulation.

The role of disclaimers​

Most AI financial tools rely on disclaimers in their terms of service to disclaim liability. These disclaimers typically say something like:

"This tool is for informational purposes only and is not a recommendation or an offer of securities. Past performance does not guarantee future results. Do not rely on this tool as your only source of financial information."

These disclaimers are legally protective (they limit the operator's liability) but do not actually tell you whether AI was used. A user who does not read the full terms of service — which is most users — will not know the content is AI-generated or that the platform disclaims responsibility for accuracy.

The effectiveness of disclaimers as a disclosure mechanism is questionable. Courts have found that disclaimers in fine print do not constitute adequate disclosure if they contradict the main message. For example, if a tool advertises "Get investment recommendations from expert analysis" in the headline, then says "this is not a recommendation, for informational purposes only" in fine print, courts might find the fine-print disclaimer inadequate.

Gaps and loopholes​

The current regulatory landscape leaves three major gaps:

Gap 1: No requirement to label content as AI-generated. Outside of specific niches (state-level synthetic-media rules, EU financial AI rules), there is no law saying "if a human reader might think this was written by a human, you must disclose that AI wrote it." A newsletter, article, or recommendation can be entirely AI-generated and published without disclosure.

Gap 2: No requirement for explainability. Even when an AI system is disclosed, there is often no requirement to explain how it works. An investment advisor might say "we use AI to manage your portfolio" but not explain the algorithm, the training data, or the validation process. A human advisor who says "I use complex methods" without explaining them would be viewed as evasive. AI advisors can get away with this.

Gap 3: Robo-advisors operate in a gray zone. Many robo-advisors claim they are not "advisors" and therefore not subject to SEC fiduciary rules. This allows them to provide AI recommendations without the governance and disclosure requirements that apply to registered advisors.

What responsible AI disclosure looks like​

A company that is transparent about its AI use includes:

  • Clear labeling. Early in the content (headline, byline, or opening), the company discloses "This analysis was generated using AI" or "This portfolio is managed by an algorithm."
  • Explanation of methods. The company explains (in plain language) what data the AI was trained on, what signal the AI is optimizing for (profit, growth, stability, etc.), and how often the AI is retrained.
  • Disclosure of cutoff dates. If the AI has a knowledge cutoff, the company says so: "Our models were trained on data through December 2023."
  • Conflict-of-interest disclosure. If the AI recommends certain securities and the company has an interest in those securities, the company discloses it.
  • Human oversight acknowledgment. The company explains what human review occurs before the AI output is published.
  • Performance history. For recommendations, the company provides historical performance (e.g., "over the past year, our recommendations have had an average return of X%").

Few companies meet all these criteria. Those that do are generally more trustworthy.

Real-world examples​

Example 1: The undisclosed robo-advisor. An investor uses a robo-advisor app that recommends a portfolio allocation of 70% stocks and 30% bonds. The app's home page features "expert analysis" and shows charts and commentary. The investor assumes a human is analyzing their situation. In the app's privacy policy (hidden 3 screens deep), it says "portfolios are generated by an algorithm." The investor has no way to know this without reading 5,000 words of legal prose. This is compliant with current U.S. regulation (probably), but ethically questionable.

Example 2: The EU-compliant fintech. A European investment platform openly discloses on its main page: "Your portfolio is managed by an AI system trained on historical market data through November 2023. Our AI optimizes for risk-adjusted returns. A human compliance officer reviews all recommendations before they are deployed. You can view our algorithm's backtest results here [link]." This disclosure is transparent and meets EU AI Act expectations.

Example 3: The deceptive newsletter startup. A financial newsletter service publishes daily market commentary. The newsletter appears to be written by a "team of market analysts" (shown in the masthead with photos). In reality, the entire newsletter is AI-generated by a single tool, and the "analysts" are AI-generated avatars. The company has not disclosed this. The FTC has begun investigating for deceptive origin representation.

FAQ​

Is my investment advisor required to tell me if they use AI?​

If they are a registered investment advisor, they are required to disclose material facts about their practices. The SEC currently does not consider AI use alone to be material (unless there is a conflict of interest). But best practices suggest disclosure anyway. If your advisor uses AI and does not mention it, ask directly.

If an AI tool has a good disclaimer, is it okay for it to hide the fact that it uses AI?​

Legally, yes — in the U.S., for most contexts. Ethically, it is questionable. A disclaimer in fine print does not count as meaningful disclosure if the main message implies human authorship.

It depends on the jurisdiction and the context. If an advisor was supposed to be registered with the SEC and was using AI without disclosing it, you might have a claim. If an AI tool genuinely disclaimed responsibility (in a way that courts find adequate), you probably do not. Consult a securities lawyer.

Are there any financial AI tools that are fully transparent about their methods?​

A few. Some academic finance tools and research platforms disclose their methods openly. But most commercial tools are less transparent, either for competitive reasons or to avoid liability.

What should I look for when evaluating an AI tool's disclosure?​

Look for: a clear statement that AI is used, the date of the training data, an explanation of what the AI optimizes for, and a human's role in the process. If these are absent, treat the tool as a commodity (perhaps useful for ideas) rather than as a trusted advisor.

Summary​

Current AI disclosure rules in the U.S. financial sector are incomplete. The SEC requires registered advisors to have governance for AI use but does not mandate disclosure to clients. The FTC prevents outright deception but does not require labeling. State and international rules are fragmentary. This means most AI-generated financial content can be published without clearly labeling it as AI-generated. You must actively seek disclosure and remain skeptical of content that does not explicitly state its origin. Responsible AI providers voluntarily disclose methods, training data, cutoff dates, and human oversight; providers that are silent on these points warrant extra skepticism.

Next​

→ AI source verification