1. Where Schwab stands on AI trading agents
Charles Schwab chief executive Rick Wurster said in mid-October 2026 that he doubts autonomous AI agents can reliably beat the market, and that the brokerage will restrict how much trading access such software receives inside client accounts. Speaking during third-quarter earnings discussions, Wurster separated AI that follows rules set by a client from AI that trades on its own, and he placed Schwab firmly on the side of human supervision.
The comments matter because AI agents are increasingly marketed as tools that can run an investor's portfolio with little or no manual input. Schwab, one of the largest retail brokerages in the United States, is signaling that it will not hand that kind of control to software by default, and that it expects clients and technology vendors to work within boundaries the firm sets.
2. What did Wurster actually say about AI agents?
Wurster's position, as relayed in coverage of the earnings discussions, is that an AI system which trades should operate under instructions defined by the user and should be watched by a person. He drew the line at fully autonomous trading, meaning a setup in which software decides what to buy and sell with no human review in the loop.
Two concerns sit behind that line. The first is liability: if an automated agent places a costly trade, it is not settled who bears the loss, whether the client, the brokerage, or the company that built the agent. The second is control. A brokerage that lets third-party software act freely inside accounts gives up some ability to predict and manage what happens there.
He was sharpest about the idea of Schwab building its own AI trading strategies for customers. Wurster called that a "fool's errand", a phrase that captures his skepticism that a machine-generated strategy can consistently outperform the broader market. The remark is notable coming from the head of a firm that serves millions of self-directed investors and could, in principle, package such a product for them.
3. How many bots are trying to get into client accounts?
Schwab says it blocks what Wurster described as tens of attempts each week by unauthorized AI agents trying to reach client accounts. No exact count was provided, so the figure should be read as an order of magnitude rather than a precise tally. It does show that automated agents are already probing brokerage systems, not merely being discussed as a future possibility.
Schwab is not alone in building defenses. JPMorgan Chase has put similar protections in place to keep unauthorized AI agents away from its clients' accounts. Taken together, the two firms suggest that large financial institutions are converging on the same posture: outside agents are not welcome to log in on a customer's behalf unless the institution has approved the arrangement.
The practical logic is straightforward. A customer's login credentials were designed to be used by that customer. When software impersonates a person to place trades, the institution loses clarity about who authorized each action, which complicates fraud detection, dispute resolution and compliance record-keeping.
4. What is Charley, Schwab's AI assistant?
Charley is the name of the conversational assistant Schwab is preparing to introduce. It is built on large language models, the same class of technology behind popular chatbots, and no launch date was given in the coverage of Wurster's remarks.
The assistant is designed to help with research, offer insights and answer account questions, such as how much money a client holds. What it cannot do is move money or execute transactions. That limitation is deliberate and follows directly from Wurster's stated worries about liability and control: Charley can inform a decision, but a human still has to make it and carry it out.
Wurster described the style of AI he prefers as one that behaves like a careful human investor. In his framing, that means researching stocks and placing limit orders, which let an investor name the price at which a purchase or sale should occur, rather than improvising in fast-moving markets. The emphasis is on disciplined, rule-bound behavior over speed or novelty.
5. Is AI driving Schwab clients away?
A common fear across the industry is that AI tools will encourage customers to move assets away from established brokerages, or to shift cash into higher-yielding places the moment a smarter option appears. Wurster said Schwab's own experience does not bear that out.
According to his comments, clients who use AI tools show more trading activity and also hold more cash. In other words, the clients using the technology are engaging more with the platform, not less. That pattern gives Schwab an argument for adopting AI gradually and on its own terms: the firm can point to evidence that careful deployment deepens client activity rather than eroding the relationship.
6. Why autonomous agents differ from ordinary trading tools
Investors have used automation for decades. Stop-loss orders, recurring purchases and rules-based rebalancing all execute without a person pressing a button, but each follows an instruction the investor wrote in advance. The behavior is predictable because the rule is fixed and visible.
An autonomous AI agent is a different proposition. It is given a goal, such as growing a portfolio, and decides for itself which securities to research, when to act and how much to risk. Its choices may vary from one day to the next and may not be easy to explain afterward. That unpredictability is exactly what makes a regulated brokerage uneasy, since the firm has duties around suitability, record-keeping and the protection of client assets that depend on knowing who decided what.
It also helps explain why Wurster's preferred model resembles a careful human researcher. A system that gathers information, presents it and lets the client choose keeps the decision with the person who bears the consequences. A system that acts alone moves that decision to software whose reasoning may be opaque.
7. The liability question behind the caution
Liability is the concern Wurster put first, and it is a genuinely open issue. If a client authorizes an outside agent and that agent buys the wrong security or trades heavily into a falling market, several parties could be implicated: the client who delegated the task, the developer who built the agent, and the brokerage that allowed the activity. Existing rules were written for human account holders and for conventional advisers, not for software that acts semi-independently.
By refusing to give outside agents free access, a brokerage narrows that uncertainty. If only approved, tightly scoped tools can act, the firm knows what each one can do and can set limits accordingly. If anything can log in and trade, the firm inherits risks it cannot easily measure.
Control matters for a second reason. Brokerages monitor accounts for unusual behavior, and a large volume of automated logins can resemble an attack. Distinguishing a legitimate client's helper from a malicious script is difficult when both present the same credentials, which is part of why blocking unauthorized agents outright is simpler than evaluating each case.
8. What comes next for AI at Schwab
The next visible step is the arrival of Charley, though the launch timing has not been specified. Because the assistant will answer questions and surface insights but not execute transactions, it gives Schwab a way to show clients the benefits of conversational AI while keeping the final action in human hands.
Wurster's comments also set expectations for the wider market. Technology companies building trading agents will find that large brokerages are unlikely to welcome unsupervised access, and that any route into client accounts will probably depend on explicit approval and clear limits. Investors who want to use AI in their financial lives can expect help with research, summaries and account questions, while the power to place trades continues to rest with them.
9. A cautious approach since 2025
Wurster has led Schwab since 2025, and his tenure has favored measured, security-first adoption of AI across the firm's services. The comments on trading agents fit that pattern. Rather than racing to launch autonomous features, Schwab has stressed assistance, information and guardrails, and it has put defensive measures against unauthorized agents in place at the same time as it prepares its own assistant.
That balance reflects the particular position of a large custodian. A brokerage holds client assets and is responsible for protecting them, so a single high-profile failure involving automated trading could damage trust far more than a slower rollout would cost in competitive terms.
10. What this means for investors, developers and brokers
For retail investors, the immediate implication is that third-party AI agents which trade freely inside an account are likely to be blocked, at least at Schwab and JPMorgan Chase. Investors who want AI help are being steered toward the assistants their brokerage provides, which are built to inform rather than act.
For AI developers, access to a brokerage is not something granted by default. Agents that attempt unauthorized entry are detected and stopped every week, so building a product that depends on logging into client accounts without the institution's approval carries a real risk of being shut out.
For the brokerage industry as a whole, the unresolved question of who pays when an autonomous agent makes an expensive mistake gives large firms a reason to keep humans involved. Until liability rules are clearer, the cautious stance Wurster described is likely to remain the default among major institutions, with Charley-style assistants serving as the sanctioned way to bring AI into the client experience.