Wall Street firms already use AI for back-office processing, compliance checks, market analysis and, in some cases, actions tied to investment decisions. The main risk is not a single model making a bad forecast. It is that many firms may lean on the same models, data or outside vendors, respond to stress in similar ways, and act faster than people can step in. Regulators have named these pathways. The official statements and reports cited in this article do not show that AI has caused a specific market crisis, and the SEC has withdrawn its 2023 proposal on conflicts of interest tied to predictive data analytics.
What “AI in finance” covers
“AI in finance” describes several different jobs, and they carry very different levels of risk. Blurring them together is the quickest way to misjudge the debate.
- Back-office and operations: processing, document handling and internal workflows. The Financial Stability Board (FSB) lists operational efficiency among the main benefits.
- Regulatory compliance: monitoring, reporting and checking activity against rules.
- Client-facing products and services: more personalized products and interactions.
- Analysis and investment-related decisions: predictions about markets, loans and credit that feed recommendations, trading or lending. Some of these systems inform a human decision; others may automate an action. The public statements cited here do not say how many do the latter.
Not every use is autonomous trading. A system that flags a suspicious payment and a model that places orders without human review carry different risks.
How firms are using AI, and what is still unknown
The FSB’s 2024 summary of an OECD-FSB roundtable reported that generative-AI use in regulated financial institutions then looked exploratory and focused mainly on operational efficiency. That describes discussions in 2024. It is not a current adoption survey.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall#1 Best Overall
In June 2024 remarks, SEC Chair Gary Gensler described AI applications ranging from call centers and claims processing to predictions about markets, loans and credit. In September 2026, SEC Commissioner Mark T. Uyeda said market participants, from retail investors to large institutions, were incorporating AI tools into investment decisions and operations.
Both statements establish that uses span a wide range. Neither says how many firms hand final decisions to a model, or how often a person reviews the output before it takes effect.
The upside is real
The FSB names operational efficiency, regulatory compliance, more personalized financial products and advanced analytics as potential benefits. For firms, faster processing and better monitoring can reduce cost and error. For clients, the stated promise is products fitted more closely to their circumstances. These gains do not remove the need to test systems before they are used or to manage conflicts of interest.
What could go wrong
The risks fall into two groups: failures inside a single firm, and pathways that link many firms together. The first group resembles the familiar risks of any model. The second is what turns AI from a technology question into a financial-stability question.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsModel and data failure
An AI system is only as good as its inputs. Data may be incomplete, erroneous, biased or out of date, and a model trained under one market condition may behave poorly under another. Poor data or weak governance can produce flawed analysis, operational disruption or bad investment outcomes. Models that are hard to interpret also make it harder for managers and supervisors to challenge a result before it is acted on. These are risks, not inevitable outcomes. The statements cited here do not measure how often such failures occur.
Conflicts of interest
An automated recommendation or client interaction can serve a firm’s incentives rather than the investor’s interests. For example, a system tuned to promote a firm’s own products, or to keep a client engaged, would pursue that goal without anyone deciding to mislead. The SEC’s 2023 proposal targeted this kind of conflict in predictive data analytics, and its status is explained below. Misstatements about how AI is used are a separate concern, covered in the regulatory section.
Correlated decisions and market stress
If many firms build on the same models, training data or data feeds, their trades, loans or prices may start to move together. Similar behavior causes little trouble in calm markets, but during a stress event it removes the diversification that normally spreads losses. The FSB identifies common models and data as a potential source of financial-stability vulnerability. It does not report that such correlation has produced a market event.
Cybersecurity, fraud and disinformation
AI can widen the attack surface because it depends on large volumes of data and often on third-party services. Generative AI can also make fraud and market disinformation more convincing or easier to produce. The FSB identifies this exposure. The official material cited here does not document a specific case in which AI-generated content was used to manipulate a market, so this should be read as a rising exposure rather than a recorded pattern.
Concentrated providers and outages
Many financial institutions may depend on a small group of providers for specialized hardware, cloud infrastructure and pretrained models. Concentration and limited substitutability mean that one provider’s disruption can become a shared problem across firms that all rely on it. The FSB’s 2025 monitoring report is the basis for this point, and the data gaps it describes are covered in the FSB section below.
What happens when AI makes the investment call
The question behind the headline is less whether a model can be wrong than what happens when its output is acted on before anyone checks it. The FSB’s concern is speed. It notes that automated strategies can respond at high speed, which could worsen a liquidity squeeze or market stress before people can intervene. A plausible sequence, which the FSB treats as a vulnerability to monitor rather than a documented chain of events, runs as follows:
Rank #3
- Many firms rely on a similar model, data feed or outside provider.
- A market shock pushes those systems toward similar signals.
- Orders, loan decisions or price changes are executed faster than staff can review them.
- Liquidity thins and volatility rises, while the people who could pause the system are still catching up.
Each link is a condition that may or may not hold in a given market. The official sources cited here do not show this sequence occurring.
Accountability is the other gap. When a model’s output is acted on, the firm still needs a named person responsible for the outcome and the authority to override the system.
Free tools Windows power users keep installed
One-click scans. No signup required.
Potential pathways versus documented events
Readers often see AI risk described in one undifferentiated pile. The table separates what the cited official material establishes from what it treats as possible.
| Claim | Status on the current record | Source and date |
|---|---|---|
| AI is used across back-office, compliance, client-facing and analytical tasks | Established as a range of uses | FSB (2024); Gensler (June 2024); Uyeda (September 2026) |
| Generative-AI use in regulated institutions was exploratory and mainly operational | Observation from 2024 discussions; not a current survey | FSB summary of OECD-FSB roundtable (2024) |
| Common models, data or providers could make firms’ behavior correlated | Identified risk pathway; not shown to have caused an event | FSB report summary (November 14, 2024) |
| Automated strategies could worsen a liquidity squeeze before humans intervene | Potential vulnerability identified by the FSB; no AI-driven flash crash documented | FSB report summary (November 14, 2024) |
| Concentrated providers could turn one disruption into a shared vulnerability | Identified risk | FSB monitoring report (2025) |
| Generative AI makes fraud and market disinformation easier to produce | Identified exposure; no specific AI-generated manipulation case verified in the material cited | FSB report summary (November 14, 2024) |
| The SEC’s 2023 predictive-data-analytics conflicts proposal | Withdrawn in June 2025; not an adopted rule | SEC withdrawal (June 2025) |
| Share of firms that let AI make final investment decisions | Not stated | Not stated in the SEC and FSB statements cited |
| Dollar losses caused by AI-directed investment decisions | Not stated | Not stated in the SEC and FSB statements cited |
Where the rules stand
The 2023 predictive-analytics proposal was withdrawn
In 2023 the SEC proposed rules on conflicts of interest linked to predictive data analytics used by broker-dealers and investment advisers. The Commission withdrew those proposals in June 2025. The SEC said it did not intend to issue final rules on them, and that any future action, if pursued, would start with a new proposal. The withdrawn proposal is not in force and should not be described as current law. Existing securities-law obligations and other applicable rules may still matter, but this article does not attempt a complete legal analysis.
AI-washing and what firms say about their AI
In a March 18, 2024 statement, Gensler told firms not to misrepresent whether or how they use AI. He said: “In essence, they should say what they’re doing, and do what they’re saying.” He also said: “AI washing, whether it’s by financial intermediaries such as investment advisers and broker dealers, or by companies raising money from the public, that AI washing may violate the securities laws.” These are a regulator chair’s statements at the time. They are not a complete description of securities law and not legal advice.
Questions from a commissioner
In March 2025 remarks, SEC Commissioner Caroline Crenshaw set out questions for regulators and market participants. They cover governance of black-box systems, compliance with legal and fiduciary duties, disclosure, investor vulnerability, and systemic or volatility risks. Her remarks state that her views are personal and not necessarily those of the Commission. They are a useful map of open questions, not SEC policy.
What the FSB asks authorities to do
The FSB’s November 14, 2024 report summary concluded: “The rapid adoption of AI in finance means that authorities should address information gaps for monitoring, assess the adequacy of current policy frameworks and enhance supervisory and regulatory capabilities.” Its 2025 monitoring report makes the same concern concrete: authorities face data gaps and a lack of standardized taxonomies as they track AI adoption and related risks. Regulators, in other words, cannot yet measure the exposure with precision, which is itself a limit on how confidently anyone can answer the headline question.
One statistic, and what it does not measure
The most prominent figure in the official material is $110 trillion. Gensler used it in June 2024 remarks to describe the scale of the capital markets the SEC oversees. It is a measure of market size. It is not an estimate of AI investment, AI adoption or AI-related losses, and it does not indicate what share of that market AI controls.
The official statements and reports cited in this article do not establish how common AI-directed investment decisions are, or what they have cost investors. If you see a precise percentage or dollar loss attributed to AI in markets, ask for the underlying data and the method used to produce it.
Questions to ask before trusting an AI-informed decision
- Does my adviser or broker tell me, in plain language, whether AI informs, recommends or executes decisions on my account?
- Is a named person responsible for those outputs, and can a human pause or override the system?
- How does the firm handle conflicts when an automated recommendation could favor its own products?
- Is the tool built in-house or supplied by an outside vendor, and what happens to my account if that vendor fails?
- What do automated processes do during a sharp market move, and who decides when they stop?
- When a firm describes its AI as accurate or superior, what evidence supports that claim beyond marketing language?
These questions apply whether the AI is doing back-office work or contributing to investment choices. The answers will often show more about a firm’s risk than any general statement about AI.
Recommended Free Tools
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
