The AI AGENT (Artificial Intelligence Access, Gatekeeper Exchange, and Nondiscriminatory Transfer) Act – backed by Senator Mark Warner of Virginia – might not become law any time soon. It faces the same challenges as many ambitious technology proposals before it: competing political priorities, difficult questions around implementation, and a rapidly evolving market that may look very different by the time legislation could realistically be enacted.

That does not mean merchants should ignore it.

In many ways, the most valuable aspect of the proposal is not whether it ultimately passes, but what it reveals about the debates beginning to emerge around agentic commerce. Until recently, most discussions about AI agents have been led by technology companies, developers and investors. Increasingly, governments are beginning to consider what a future economy may look like when consumers delegate decisions and online activity to autonomous agents. The AI AGENT Act provides one of the clearest indications yet of how policymakers are starting to approach that question.

Moving Beyond AI and Towards Agentic Commerce

At its core, the proposed legislation (as launched here on June 29th 2026) would create a legal framework allowing consumers to authorize AI agents to act on their behalf across large online platforms. Companies with more than 50 million U.S. users would be required to make accessible interfaces available so that authorized third-party agents could perform many of the same actions as a human user. The concept effectively introduces an open access model for AI agents operating across online platforms, commerce, financial services and digital ecosystems – sort of like how open banking works today.

What makes the proposal particularly interesting is that it is fundamentally a competition and consumer choice discussion rather than an AI discussion. The legislation assumes that consumers should have the right to choose whichever agent they want to represent them online, and that large platforms should not be able to unfairly restrict that choice. Whether or not the bill progresses, it offers a clear window into how policymakers are beginning to think about the future relationship between consumers, platforms and autonomous software.

A Debate Payments Teams Have Heard Before

For those working in payments, this line of thinking should feel familiar. The industry has spent years debating network access, routing rights, wallet competition and consumer choice. The underlying question is often the same: when multiple parties have interests in a transaction, who ultimately gets to decide how it happens?

The AI AGENT Act appears to apply a similar philosophy to agentic commerce. If consumers choose to delegate decisions to an AI agent, should platforms be able to block that agent, favor their own competing solutions or limit access to information? The parallels with historical payments debates around open banking, debit routing and payment method acceptance are difficult to ignore.

Looking specifically at commerce, the proposal raises important questions around merchant acceptance. If consumers configure agents to favor particular funding methods, rewards structures or payment products, how much influence will merchants retain over payment selection? If an agent is instructed to only pay using a method that a merchant does not accept, breadth of acceptance could become increasingly important. While these scenarios remain speculative, they illustrate how agentic commerce could eventually reshape assumptions around checkout optimization and payment choice.

The Immediate Priority Is Visibility

While these long-term implications are interesting, merchant payments teams should be focused on something far more practical today: visibility.

Most organizations are still debating whether agentic commerce will become meaningful in one year, five years or ten. In reality, the timing matters less than many people think. As soon as agent-driven activity starts appearing within a merchant environment, it should be identifiable, measurable and analyzed separately from traditional customer activity.

The payments industry already provides a useful model. Consumer Initiated Transactions (CITs) and Merchant Initiated Transactions (MITs) are almost always monitored separately because they exhibit different behaviors, performance characteristics and risk profiles. Agent Initiated Transactions (AITs) should be viewed in the same way. Combining them with broader transaction populations may obscure valuable insights and make it difficult to understand how autonomous systems behave.

The merchants that establish this visibility early will be able to understand how agents discover products, evaluate offers, select payment methods and move through purchase journeys. More importantly, they will build an evidence base before agentic commerce reaches material scale.

Fraud May Become the Biggest Challenge

Much of today’s discussion around agentic commerce focuses on convenience, automation and conversion. These benefits are real, but they are only one side of the equation. Any technology that allows legitimate activity to operate at scale also creates opportunities for malicious actors to do the same.

The AI AGENT Act acknowledges this challenge through discussions around liability, misuse and fraud, including proposals to explore frameworks for determining responsibility when agents make mistakes or cause harm.

For merchants, however, the operational challenge arrives long before any formal liability framework. Human fraud attacks already exploit automation extensively and agentic fraud may simply make those attacks faster, cheaper and more scalable. Autonomous systems can continuously test offers, probe merchant controls, evaluate responses and adapt behaviour in ways that traditional fraud operations cannot.

As a result, the challenge is not about simply enabling agentic commerce but identifying and rejecting bad agentic traffic at scale. The ability to distinguish legitimate consumer agents from malicious automated activity could become a core capability for payments and fraud teams over the coming years.

Influencing Algorithms Rather Than Humans

There is also a less obvious implication for payments optimization. Today’s eCommerce experiences are designed largely around influencing human behavior. Merchants invest heavily in checkout design, payment placement, promotions and customer experience improvements that encourage particular outcomes. Autonomous agents are unlikely to respond to these signals in the same way.

As discussed within our monthly Agentic Commerce Working Group, part of our work with the Merchant Payments Leadership Council (MPL Council), an AI agent may need only a microscopic economic advantage to select one option over another consistently. A difference that would be imperceptible to a human customer could become highly meaningful when replicated across millions of automated decisions.

In this environment, optimization becomes increasingly mathematical. Payment economics, rewards structures, transaction success rates, acceptance breadth and cost efficiency may carry far more weight than visual design or emotional messaging. Merchants may find themselves competing for algorithmic preference rather than for consumer preference.

What Merchants Should Do Next

The AI AGENT Act should not be viewed as evidence that agentic commerce has arrived. Rather, it is evidence that governments are beginning to think seriously about what happens if it does. 

For merchant payments teams, the appropriate response is neither panic nor complacency. Agentic commerce remains early, standards remain immature and long-term outcomes remain uncertain. However, the merchants that start building visibility today will be better positioned regardless of how quickly adoption occurs. 

The first steps are straightforward: 

• Identify agentic traffic when it appears.  

• Create dedicated reporting.  

• Analyze AITs separately from traditional customer activity.  

• Understand how autonomous systems interact with your payment methods, fraud controls and customer journeys.  

The merchants that take these steps now will be in a much stronger position as agent-driven commerce evolves from experimentation into a meaningful channel. 

Whether the AI AGENT Act ever becomes law is ultimately a secondary consideration. The more important takeaway is that policymakers are beginning to ask the same questions that merchants will eventually have to answer. When software starts acting as the customer, what changes? 

For payments teams, the answer is likely to be far more than most organizations currently realize.