From public attestations to PKI, industry explores how to stop secret agents

It’s 2026; do you know who your agents are? This is the question of the moment, as the agentic AI explosion of late 2025 meets the question of how to trust the machines we call AI. It is, to quote a recent analysis from PYMNTS, the “KYA Moment.”

Know Your Agent used to mean taking your realtor’s business card. In the world of neural networks and large language models that enable dedicated bots to perform transactions on a human’s behalf, it refers to “the trust layer that makes delegated, machine-initiated commerce legible to networks, issuers and merchants before fraud systems intervene.” Like Know Your Customer (KYC) protocols, KYA establishes the identity attributes, authorizations and permissions of an entity – in this case, an AI agent rather than a human customer.

“In practical terms,” says PYMTS, “it turns delegated intent into something that can be verified rather than inferred.”

Chain enables reputational trust through public attestations: Billions

A blog from Billions looks at the trust gap in agentic AI systems, and how its blockchain-based approach can address it.

“When an AI agent behaves incorrectly, there is no clear way to understand who built it, who controls it, or whether it can be trusted at all,” it says. “As AI adoption grows across business, finance, and public systems, this gap becomes a real risk, not just a technical limitation.”

An AI agent, it argues, needs a “real” identity, by which it means an encrypted Decentralized Identifier (DID) about which others on the chain can make public attestations: “anyone can verify what is known about an agent without trusting the agent itself.” Trust is built from relationships, rather than permissions.

Billions provides an Agent JS SDK that allows AI agents to generate and manage DIDs, perform authentication, and interact with the public attestation registry.

Decentralized identity cuts out need for hierarchies: Kaliya Young

A recent episode of the Knowledge Hub podcast features insights from Kaliya Young, a 20-year veteran in the decentralized identity sector – making her among the first to recognize its potential. Young breaks down the concept of namespaces, in an explanation of why identity becoming decentralized, so to speak, is such a big deal.

“We have a world of centralized identifier management,” she says. “So you have phone numbers managed in a global phone number system, DNS managed in your domain names and IP addresses managed globally by ICANN and IANA. And then you have private name spaces: your Gmail address is managed by Google, your Proton mail address is managed by Proton Mail. You have names within other people’s context.”

“Those are all different versions of centralized identity systems, requiring you to interact with a hierarchy.”

Decentralized identity, on the other hand, “gives us a way to have globally resolvable identifiers –  meaning really long numbers that have with them associated key material to communicate with the endpoint.” The key is embedded, instead of having to be signed from a certificate authority.

The significance in the context of AI agents is that globally resolvable decentralized public key infrastructure enables the provisioning of identifiers without the need to access a centralized registry. Anything can be provisioned – objects, organizations, or agents.

“As we enter this era of AI agents, it’s going to be critical to understand which agent is trying to access which service, who’s actually asking the agent to act on their behalf, what capacities do they have, do they trust the agents. There’s a whole range of questions that come up as we look into the future of much more complex identity and credentialing requirements that agents need.”


Source: biometricupdate.com