The Missing Link in Visa’s AI Commerce Vision: Identity, The Identity Crisis at the Heart of AI-Powered Commerce

By Samuel Pearton

Visa's recent unveiling of its vision for AI agents reshaping commerce marks a pivotal moment in how we interact with the digital economy. According to Visa, the future will involve AI-powered agents that not only help consumers discover products but also autonomously make purchases on their behalf. This is more than a feature update to how we shop; it represents a foundational shift in who or what is transacting.

But before this vision can become reality at scale, we must confront a challenge more fundamental than product discovery, recommendation algorithms, or payment processing APIs. We must solve the identity problem.

The Scope of the Fraud Problem

The United States saw over $12.5 billion in consumer fraud losses in 2024 alone, a 25% increase from the previous year. Of this, credit card fraud remains the most commonly reported form of identity theft, accounting for over 326,000 reports in 2024. A staggering 73% of these fraud losses now stem from online, card-not-present transactions where neither the cardholder nor their card physically appears at the point of sale.

This is not just a statistic. It is a signal. The explosion of card-not-present fraud points to the root issue: identity.

Today’s payment systems were designed for human cardholders swiping plastic at terminals. But that model is already obsolete. Most fraud now takes place when identity is abstracted into a line of code or a digital profile. Tomorrow’s agents, AI-powered and autonomous, will not carry wallets. They will carry credentials. And we are not yet ready for that shift.

AI Agents, a New Vehicle for Fraud

As Visa points out, AI agents will soon be embedded across platforms, apps, and devices. These agents will shop for us, negotiate for us, and pay for us. But if we think bad actors aren’t already plotting how to take advantage of this shift then we are severely underestimating them.

AI agents introduce new attack surfaces. Consider a rogue agent spoofing the behavior of a legitimate one, interacting with merchants, issuing fraudulent purchase requests, or misrepresenting user intent. Now add scale. These agents will operate constantly across regions and platforms, handling sensitive information and initiating transactions potentially worth thousands of dollars. Without a verifiable identity layer, we are not solving fraud. We are scaling it.

Why Identity Is the Cornerstone

What is missing is a mechanism to prove that an AI agent is:

1. Legitimately representing a known individual.
2. Running the correct, approved model and logic.
3. Acting within its authorized boundaries.

None of this can be assumed. It must be proven.

In today’s systems, verification often relies on trust such as API keys, client-side tokens, or black-box service providers. But trust is not proof. AI agents need cryptographic guarantees. They need to be verifiably who they say they are and provably doing what they say they are doing.

The Role of zkML

This is where zero-knowledge machine learning (zkML) comes in. zkML enables agents to embed a cryptographic “watermark” into their decision-making process. This watermark can prove that a specific AI model was used on specific inputs to generate a specific output without revealing the model itself or the input data. In essence, it turns AI behavior into a verifiable claim.

With zkML, we can:

• Assign a verifiable identity to each AI agent, rooted in the specific logic and behavior it runs.
• Enable merchants and payment processors to verify that an agent is authorized to act on behalf of a given user.
• Protect user privacy while providing high-assurance authentication and auditability.

Instead of asking, "Can I trust this AI agent?" businesses and platforms can ask, "Can I verify this agent’s identity and behavior?" That shift from trust to proof is how we secure AI-native commerce.

Towards a Verifiable Commerce Infrastructure

Visa’s roadmap is ambitious and in many ways inevitable. AI agents will become the norm, not the novelty. But the infrastructure they rely on must be re-architected with verifiability at the core. Payments, authorization, dispute resolution, and fraud detection all need to evolve to accommodate a new kind of actor, one that is fast, intelligent, and invisible.

The good news is that we do not have to start from scratch. Cryptographic primitives like zero-knowledge proofs have matured. zkML is advancing rapidly. And the shift toward decentralized, privacy-preserving identity is already underway. What is required now is convergence: of technology, standards, and stakeholder will.

Final Thoughts

Visa is right to push the frontier. But as AI agents start to navigate the economy on our behalf, they must be equipped with the same or better credentials than the people they represent. Verifiability is not a luxury. It is a prerequisite.

Before we solve the AI agent fraud problem, we must solve the identity problem. And that identity must be rooted in proof.

Only then can we unlock the true potential of autonomous commerce: secure, scalable, and trustworthy by design.

Author Bio

Samuel Pearton is the Chief Marketing Officer at Polyhedra, driving the future of intelligence through its pioneering, high-performance technology in EXPchain—the everything chain for AI. Drawing on decades of experience in tech, global marketing, and cross-cultural social commerce, Samuel understands that trust, scalability, and verifiability are essential to AI and blockchain.. Before officially joining Polyhedra’s executive team in October 2024, he played a key advisory role as the company secured $20 million in strategic funding at a $1 billion valuation. Prior to Polyhedra, Samuel founded PressPlayGlobal, a social commerce and engagement platform that connected athletes and celebrities—including Stephen Curry and other leading global brands—with China’s largest consumer fan market.