Delegated identity stack emerges for AI agents to authenticate, act and pay

In movies, being a secret agent is a cool way to immerse yourself in espionage and intrigue; in authentication, it’s a problem you don’t want to have. AI agents are performing tasks on behalf of humans, and while it’s important to prove an agent is authorized to act as your proxy, it’s becoming equally important to do so without the agent – for authentication purposes – becoming you.

A trio of new developments point toward an identity and trust stack for delegated agency. They come from 1Password, Dai Nippon Printing (DNP) and the Linux Foundation, respectively, and they address the same problem from different angles.

1Password lets users tell Claude what’s allowed

A blog from 1Password looks at its solution for Anthropic’s AI large language model (LLM), Claude. “Once an agent can click, buy, update, and submit for you,” it says, “the key question becomes: what identity is it acting under, and what access should it get?”

Which is to say, once Claude reaches a login page, there is a calculation to be made. “Do you give the agent your password, or stop and do the task yourself?” In 1Password’s view, “neither is the future we should build toward.”

The company’s product for Claude is built on a zero-exposure architecture, meaning Claude can complete browser tasks that require logins and one-time passcodes – “but the credentials never enter the model or its memory. 1Password stays the source of truth for the secret, and access is granted only at runtime.”

In effect, 1Password acts as an authority to step in and authenticate a user without disclosing unnecessary information. Per the blog, when Claude needs to sign in, “1Password shows the user which credential is being requested and why.”

“After user-consented biometric approval, 1Password injects the credential directly into the page. Claude never sees the vault item, password, or one-time code. Access is scoped to the current task and ends when the task is complete. After autofill, 1Password checks that secrets were not exposed on the page.”

Nancy Wang, CTO of 1Password, emphasizes that the answer isn’t handing agents your secrets. “It is to let a user give an agent permission to use a credential without letting the agent see it. Claude knows it used your login; it does not need the password or one-time code in its context. That distinction is where trust in agents starts and the foundation we’re building with Anthropic.”

DNP system uses VCs to support agentic commerce

Dai Nippon Printing has begun providing a digital identity management function for AI agents when making proxy purchases in online shopping, according to an announcement auto-translated from Japanese. The system is based on DNP’s distributed ID management platform, CATRINA, and leverages verifiable credentials  to “create an environment where only AI agents authorized by the user can conduct transactions based on the user’s will and authority.”

With agentic commerce, the firm says, “a new challenge has emerged in confirming and verifying the user to determine ‘whose proxy’ and ‘with what authority’ the AI ​​agent is acting, and realizing legitimate proxy purchases.” DNP’s solution is to leverage its identity infrastructure to express a user’s delegated authority to the agent.

“By extending the identity verification and access control technologies cultivated in CATRINA to AI agents, we provide a foundation that supports safe, transparent, and fair online transactions required in an era where humans and AI agents collaborate.”

Linux launches x402 open protocol for HTTP payments

A release from the Linux Foundation announces the operational launch of the x402 Foundation, a new open-governance body created by Coinbase to steward the x402 protocol, an open standard for internet-native payments over HTTP. Per the release, x402 “embeds secure payment capabilities directly into web interactions so that AI agents, APIs, and applications can send and receive payments as seamlessly as they exchange data, with support for payment types ranging from traditional cards to stablecoins.”

The Linux Foundation will govern the x402 Foundation, allowing developers, financial institutions, cloud providers and other community members to collaboratively shape the development of the protocol.

“AI agents and automated systems are becoming active participants in the global economy, yet they have lacked a native, secure way to transact,” says Jim Zemlin, CEO of the Linux Foundation.

“The operational launch of the x402 Foundation marks a vital milestone in establishing an open, community-governed standard for payments over HTTP. By bringing together leading companies across finance, technology and more, we’re ensuring that the payment layer of the internet remains neutral, highly interoperable and ready to support digital commerce.”

The Foundation already boasts 40 member organizations, including premier members Adyen, Amazon Web Services (AWS), American Express, Circle, Cloudflare, Coinbase, Fiserv, Google, Mastercard, Monad Foundation, MoonPay, Ripple, Shopify, Solana Foundation, Stellar Development Foundation, Stripe and Visa.

In the architecture of agentic commerce, if DNP addresses an agent’s authority to buy, and 1Password addresses its authority to access the account needed to buy, x402 addresses the transaction itself. Together, these developments illustrate how emergent authentication efforts can harness the power of AI without sacrificing customer agency, privacy or security, by separating the person, the agent, the authority delegated to it, the credentials it may use and the transactions it can execute.

Article Topics

1Password  |  agentic commerce  |  AI agents  |  authentication  |  Dai Nippon Printing (DNP)  |  digital identity  |  identity governance  |  Linux Foundation