Agentic commerce AI verification is rapidly becoming a focal point as artificial intelligence systems take on larger roles in searching, negotiating and even completing transactions on behalf of users.

In this emerging model, AI agents don’t just recommend products. They compare prices, negotiate terms and, in some cases, execute payments. Payment approvals, merchant onboarding and recurring billing are increasingly embedded into automated decision loops.

That shift is changing how identity controls are applied — and who, or what, must be verified.

Agentic Commerce AI verification and the Rise of Know Your Human

As AI systems become more capable of mimicking human behavior, traditional identity frameworks are being tested. Know Your Customer (KYC) verifies that a person exists and meets regulatory requirements. Know Your Business (KYB) does the same for companies.

But in agent-driven environments, software is initiating actions. That introduces a new layer of scrutiny.

The Wall Street Journal reported on Feb. 26 that Sam Altman’s human verification startup, Tools for Humanity, is working with major consumer brands to promote its World ID product as a proof-of-human solution. The goal is to distinguish real individuals from bots in digital interactions.

This push aligns with the broader evolution from KYC and KYB toward Know Your Agent (KYA), a framework articulated by Trulioo and discussed in recent months by PYMNTS. KYA focuses on verifying that an agent is legitimately acting on behalf of a customer or business.

Extending that further introduces the concept of Know Your Human KYH identity fraud AI agents safeguards. In this model, verification ensures that a real person authorized an action and that the AI agent remains within its delegated authority.

As AI agents gain autonomy in financial workflows, independent verification of human intent is increasingly viewed as essential.

The Cost of Gaps in Digital Identity

The commercial risks tied to weak controls are already measurable.

According to the PYMNTS Intelligence report, conducted in collaboration with Trulioo and titled “The Hidden Costs of ‘Good Enough’: Identity Verification in the Age of Bots and Agents,” companies are losing an average of 3.1% of annual revenue due to digital identity gaps. These losses stem from fraud, misidentification and compliance failures.

Across the 350 global companies surveyed, that amounts to $94.9 billion in annual losses.

The study found that 56.3% of firms face threats related to bots or agents. Meanwhile, 58.6% reported struggling with bot-fueled fraud, and more than four in 10 experienced incidents or losses tied to adversarial bots and agents.

Despite those figures, 96.3% of respondents said they were confident in their ability to detect harmful bots. That gap between confidence and actual exposure highlights limitations in existing onboarding checks and verification frameworks, particularly as automation scales.

As agentic systems become more embedded in payment flows, AI agent payment approval fraud risk becomes a more immediate operational concern. When an automated system approves or initiates transactions, the verification question shifts from “Who is the customer?” to “Was this action genuinely authorized by a human?”

Liability and the Next Layer of Compliance

In an interview with PYMNTS, Trulioo Chief Product Officer Zac Cohen described Know Your Agent as a natural extension of existing compliance standards.

“We want to understand who the agents are,” Cohen said. “We want to make sure that they carry the instructions and the prompts of the individual, specifically of how they should be.”

Cohen also pointed to liability as a central barrier to broader adoption of agent-driven commerce.

“That’s the big sticking point for a lot of these transactions to really take off,” he said.

As agentic commerce expands, the shift toward agentic commerce AI verification reflects a broader recalibration of digital trust. Verification is no longer just about onboarding a user or business. It is increasingly about confirming that automated systems act within human-approved boundaries.

The evolution from KYC to KYA and toward Know Your Human frameworks signals how identity standards are adapting to a world where software is no longer just assisting transactions — it is executing them.