1Password aims to cut standing access risks for AI agents

UPDATED 09:00 EDT / JULY 28 2026 SECURITY 1Password aims to cut standing access risks for AI agents by Duncan Riley Identity security company 1Password today launched 1Password Privileged Access, a product that gives engineers and AI agents a single task’s worth of access to critical infrastructure and nothing that outlasts it. The product pushes 1Password’s Unified Access platform into privileged access management, a category the company entered in June with the acquisition of Apono Inc. Privileged Access is built on the startup’s just-in-time provisioning technology. Permissions are written directly into the target system’s native policy layer across cloud environments, databases and developer infrastructure. Existing tooling stays where it is. An agent never needs the underlying credential. Standing access is the target. It accumulates quietly as engineers, service accounts and identity and access management roles pick up permissions for one task and keep them long after the task ends. AI agents widen the blast radius. They inherit the privileges of whoever deploys them, or hold on to credentials they were handed. Recent 1Password research found that 40% of developers grant agents persistent access to systems or credentials. “Most organizations have more standing access in their environments than they can see or justify,” said Chief Executive David Faugno. “That invisible access creates exposure and too often, companies discover it only after an attacker does.” Privileged Access tracks identities and permissions across cloud, database and Kubernetes environments to surface access that should be reclaimed or right-sized. An account or privilege is created only when someone asks for it, and it covers the task at hand. When the session ends, it is gone. Every request and approval is logged with full attribution. That covers audit evidence for Service Organization Control 2, the Health Insurance Portability and Accountability Act and the Payment Card Industry Data Security Standard. ISO 27001 and the European Union’s General Data Protection Regulation are handled the same way. Low-risk requests clear automatically under policy. Higher-risk ones go to a reviewer through tools such as PagerDuty, Slack, Microsoft Teams or Jira. 1Password also moved 1Password Credential Broker into public preview for GitHub Actions. The tool debuted in private beta in June. It issues credentials scoped to an individual workflow run, keeping long-lived static secrets out of pipeline configurations. Before releasing anything, the broker verifies that the requesting identity is trusted and logs the delivery. Ben De St. Paer-Gotch, director of product management at GitHub Inc., said in the announcement that engineering teams can now strip secrets out of pipeline configurations while still giving GitHub Actions what it needs to build and ship software. Three capabilities aimed at developer environments also reached general availability in 1Password Enterprise Password Manager. Developer Watchtower finds exposed credentials in local .env files and prompts developers to move them into 1Password, while giving administrators a view of credential risk on those machines. 1Password Environments lets developers import existing .env files into the vault and reach secrets through the company’s Model Context Protocol server, keeping them off disk and out of model context. Credential Governance gives administrators a central view of company-owned credentials sitting in employee and shared vaults, the ability to take ownership of them and control over how they are accessed over time. The company’s enterprise vault holds more than 1.5 billion credentials and secrets.