The landscape of mortgage servicing and consumer financial services is on the precipice of a fundamental shift. While it is not yet a daily occurrence for a mortgage servicer to receive a call from a customer’s autonomous AI agent, the technological and regulatory infrastructure required to make this a reality is converging with remarkable speed. As artificial intelligence moves from being a tool for human assistance to an "agentic" entity capable of making decisions and initiating actions, the financial sector faces a new set of challenges regarding trust, accountability, and security.
According to a recent International Monetary Fund (IMF) staff note, agentic AI is already beginning to influence the world of payments, steering the industry toward a future defined by agent-mediated decisions. This trend is further supported by a 2026 Cloud Security Alliance survey, which revealed that 85% of respondents in the financial services sector expect AI agents to begin initiating payments on behalf of consumers in the near future. For mortgage servicers, this means the day is fast approaching when a borrower’s AI might call to negotiate a repayment plan, dispute a fee, or request a loan modification.
Organizations that wait until these interactions become routine before developing a response strategy will find themselves in a precarious position. Writing policy under the pressure of a crisis—after a system failure or a security breach—is rarely effective. The institutions that take the time to think through these implications now will be the ones to set the industry standards, rather than being forced to react to the standards established by others.
How I got here
The urgency of this issue became clear to me following a specific incident involving a customer-facing AI system for which I was responsible. Over the course of several weeks, the system’s performance began to quietly and subtly degrade. It wasn’t a catastrophic failure that triggered immediate alarms; instead, it was a slow erosion of quality that went unnoticed by human supervisors for some time.
This experience led to an uncomfortable but necessary realization: If an organization’s own AI agent—a system built, managed, and monitored by internal experts—can lose its efficacy without anyone noticing, what level of confidence can we have in an agent built by a third party? When a customer’s AI agent calls a servicer, the servicer has zero visibility into that agent’s underlying logic, its training data, or its current state of "fitness." Relying on an external black box to represent a borrower’s best interests introduces a layer of risk that the industry is currently ill-equipped to handle.
Why the existing authentication playbook doesn’t transfer
The traditional method for verifying a caller’s identity in the mortgage industry is well-established. Servicers use a series of questions to confirm that the person on the line is either the borrower or a properly authorized representative. This process relies on "knowledge-based authentication," asking for information that only the real borrower would know. However, this entire framework is built on the assumption that the entity on the other end of the line is a human being.
The playbook fails to account for a more complex scenario: what happens when the entity correctly answering every verification question is not a person, but an AI agent acting on a person’s behalf?
The instinctive reaction from many in the industry is to simply apply the same verification standards. If the AI knows the Social Security number, the property address, and the last payment amount, it should be treated like any other authorized caller. But this approach misses a critical distinction. A human representative, once authorized and trained, does not typically undergo a "silent degradation" of their faculties over time. They do not suddenly lose the ability to understand context or start hallucinating facts. An AI agent, however, can suffer from model drift or software updates that fundamentally change its behavior and reliability. A framework designed to verify a human’s identity is insufficient when the caller is a machine that can "break" in ways a human cannot.
Three distinct questions, not one
To address the complexities of AI-to-AI communication, it is helpful to deconstruct the broad concept of "AI agent trust" into three specific, manageable questions.
The first is disclosure. When an AI agent calls a servicer on behalf of a customer, does it have a legal and ethical obligation to state clearly that it is an artificial intelligence? There is a strong argument that the answer must be yes. We already have established rules requiring the disclosure of call recording or the use of automated dialing systems. Transparency is the bedrock of financial interactions; a servicer must know whether they are negotiating with a person or an algorithm to ensure the conversation is handled through the appropriate protocols.
The second is authentication. Even if an agent identifies itself honestly, how does the servicer know that this specific agent was actually authorized by the customer? The payments industry is already working on technical solutions for this. Visa’s "Intelligent Commerce" and Mastercard’s "Agent Pay" are frameworks that allow AI agents to prove their authorization and operate within strictly defined spending limits. However, no such standardized framework currently exists for the more complex world of loan servicing, where the stakes involve long-term debt and home ownership rather than simple retail transactions.
The third question, and perhaps the most overlooked, is ongoing fitness. This is where my own experience with model drift becomes highly relevant. Even if an agent is properly authenticated today, is it still capable of representing the customer’s interests effectively six months from now? AI models are not static; they change based on new data and environment shifts. An agent authorized last year may have quietly degraded in quality, yet most current credentialing frameworks only focus on the initial "handshake" of authorization. They remain silent on the question of whether the agent is still "fit" to perform its duties.
Where this leaves the asymmetry question
A common counterargument in the tech community is one of symmetry: If a mortgage servicer is allowed to use its own AI agent to speak with customers, why shouldn’t the customer be allowed to use an AI agent to speak back on equal terms?
The answer to this lies in accountability, not in a fear of the technology itself. When a mortgage servicer deploys an AI agent, there is a clear line of responsibility. The servicer owns the system, manages the monitoring infrastructure, and carries the legal liability if the system fails or provides incorrect information. In a corporate environment, a system cannot "own" an outcome; only a person can. The servicer has designated individuals who are ultimately responsible for the agent’s actions.
Most individual consumers do not have this level of governance. If a consumer’s AI agent fails, the consumer bears the cost, but the servicer is left dealing with the fallout of an unreliable negotiation or an unauthorized request. There is a fundamental difference between "personal acceleration"—using AI to make life a little easier—and "mission-critical automation."
When an AI agent is used to negotiate a loan modification or dispute a charge, it is performing a mission-critical task. Treating a consumer’s unmonitored AI agent as the equivalent of a highly governed corporate AI system ignores the inherent risks and the lack of accountability on the consumer side. A recent Forrester survey highlighted this gap in trust, finding that only about 24% of U.S. online adults trust an AI agent to act on their behalf even for routine purchases, let alone complex financial restructuring.
Furthermore, current regulations such as Regulation E in the U.S. place significant responsibility on consumers once they authorize a third party. Similarly, the EU’s Payment Services Directive (PSD2) requires clear human authorization for payment orders, with no existing mechanism to treat an AI agent as a human payer’s legal equal. The technology is outstripping the legal frameworks intended to govern it.
How common is this, really?
To be clear, we have not yet seen a widespread trend of AI agents calling mortgage servicers to negotiate complex loan modifications. It is not yet a documented industry standard. However, all the signs in adjacent domains point to this being the inevitable next step. Banks and payment networks are already piloting agent-initiated transactions, and regulators are beginning to draft the rules that will govern these interactions.
The lead time for building sound policy in the mortgage industry is measured in months, not years. If the industry waits until an incident occurs—such as an AI agent inadvertently agreeing to terms that harm a borrower, or a servicer’s system being overwhelmed by automated AI "callers"—it will be too late to develop a thoughtful, proactive strategy.
What I’d propose
Given the current lack of reliable ways to authenticate customer-authorized AI agents or monitor their ongoing fitness, mortgage servicers should adopt a stance of deliberate caution. Until industry-wide standards emerge, I propose a "practical bridge" to manage these risks.
First, if a servicing system suspects it is communicating with an AI agent rather than a human customer, the protocol should be to route the conversation to a live human representative. This ensures that mission-critical decisions are overseen by people who can be held accountable.
Second, servicers must begin developing explicit policies for identifying, authenticating, and evaluating AI agents. These rules should be integrated into existing AI, IT, or customer care governance programs. The policies should clearly define what an agent is permitted to do and, more importantly, what it is prohibited from doing.
These measures should not be viewed as a permanent rejection of AI-to-AI interaction. Instead, they serve as a necessary safeguard while the industry works toward stronger, more transparent standards. The goal is to address the risks of agentic AI deliberately and strategically, before a crisis forces the industry to improvise in the dark. By establishing these boundaries now, mortgage servicers can protect both their operations and their customers as we enter this new era of automated finance.
