1Password launches Unified Access to secure AI agents' logins, tokens, and secrets

1Password has unveiled Unified Access, a platform meant to close a new security gap that opens up as companies bring AI agents into their workflows. If agents need logins, API keys, SSH keys, and other secrets to do real work, companies have to know where that data lives, who uses it, and on what terms.

Where the risk is growing

The problem is that classic IAM and SSO systems mostly control the moment a user signs in, not what happens to credentials afterward. AI agents work differently: they run in browsers, IDEs, local scripts, and automations, and they call SaaS services and APIs. To do that, they get access to secrets. Those keys and tokens often sit in .env files, configs, shared documents, or directly on developers' laptops.

The result is a blind spot between "the user signed in" and "the agent started acting on their behalf." That is the gap 1Password is targeting. The company announced Unified Access on March 17, 2026, and pitches it as a single access-management layer for people, AI agents, and machine identities.

1Password stresses that the risk is no longer theoretical: agents are starting to operate inside production environments, which means they can reach real data, internal services, and critical workflows. The more teams experiment with AI automation, the faster untracked secrets, temporary workarounds, and excessive permissions pile up.

"Agents are already operating inside real production environments,"

1Password said.

How the platform works

Unified Access is built around three steps: discover, secure, and audit. The first two are available now; full action auditing is due later. At launch, the company also opened up Unified Access Pro.

First, the system shows which AI tools and local agents are actually in use across the company and finds exposed secrets in developer environments. Teams can then quickly move that data into a single encrypted vault and apply shared access policies for employees, agents, and machine processes. The company says the underlying technology is already used by more than 180,000 organizations and protects over 1.3 billion credentials and secrets.

• Discovery of AI tools and local agents on devices and in browsers
• Detection of unprotected .env files, API keys, and unencrypted SSH keys
• Migration of secrets into a single vault with shared policies
• Access control for risky and shared accounts
• Unified auditing of who used which secrets, when, and on whose behalf — this part has been announced but has not launched yet

The next step is issuing secrets not "forever" but exactly when they're needed. 1Password is betting on a just-in-time access model: an agent or automation receives credentials in the context of a specific task instead of storing a long-lived token wherever it happens to land.

Later this year, the company promises to add scoped credentials for agentic and machine processes at runtime. That should cut the number of standing permissions, reduce secret sprawl, and make it easier to control exactly what an agent can do.

Launch partners

Unified Access isn't launching as a standalone product but as part of the AI and developer ecosystem. 1Password's announced launch partners include Anthropic and OpenAI, along with Cursor, GitHub, and Vercel.

With Anthropic, the work covers integration with the Claude browser extension, Cowork, and Claude Code, so Claude can sign in to services with the user's consent and without the password being handed directly to the model. With OpenAI, the company says it is collaborating on the secure use of local 1Password vault items and on developer IDE scenarios.

Separately, 1Password is moving into the infrastructure and browser layer, where the risk is especially high because the agent is already acting on the user's behalf. The list includes Perplexity, Browserbase, Anchor Browser, KERNEL, Natoma, Runlayer, and Commvault.

The logic is clear: the closer AI agents get to the browser, CI/CD, and corporate data, the more important it becomes not just to store secrets but to control when they're issued and keep a clear trail of actions across the entire chain. That's what 1Password is trying to build a new standard for agent security on.

What it means

The market is shifting fast from "how do we give employees a convenient login" to "how do we manage access shared by people, bots, and AI agents." 1Password is trying to claim this layer before its competitors do — not as yet another password manager, but as a control plane for secrets, tokens, and agent actions.

If AI automation really does go into production at scale, demand for systems like this will only grow, especially in the enterprise.