1Password Privileged Access targets over-permissioned AI agents
1Password launched Privileged Access to issue task-scoped permissions for engineers and AI agents across cloud, database and dev systems.
By Wei-Lin Zhao · AI Correspondent
· 3 min read
1Password launched 1Password Privileged Access, a privileged access management product that gives engineers and AI agents temporary access for specific tasks instead of long-running permissions. The move takes 1Password further into enterprise infrastructure security after its June acquisition of just-in-time access startup Apono Inc., and it addresses a problem that has become sharper as companies put AI agents near production systems.
The company did not disclose pricing, customer count for the new product or revenue tied to the launch. 1Password said the product is part of its Unified Access platform and is built on Apono’s just-in-time provisioning technology.
What is 1Password Privileged Access?
1Password Privileged Access is designed to create access only when a user or agent requests it, limit that access to the work being performed and remove it when the session ends. The product writes permissions into the native policy controls of the target system, including cloud environments, databases and developer infrastructure, rather than requiring teams to replace their existing tooling.
For AI agents, the practical claim is that the agent does not need to receive or store the underlying credential. That distinction matters for security teams because agents often inherit the access of the person or system that deploys them, which can turn a narrow automation task into a broader exposure if credentials persist.
1Password said its own research found that 40% of developers give agents persistent access to systems or credentials. The company is positioning the product against standing access: permissions granted for one job that remain available after that job is complete. In large engineering organizations, those permissions can be spread across human users, service accounts and identity and access management roles.
Chief Executive David Faugno said in the company’s announcement that many organizations have more standing access than they can see or justify, and that this hidden access is often discovered after an attacker finds it. That is a familiar vendor argument in privileged access management, though the AI-agent angle gives it a newer operational target.
How it handles approvals and audits
1Password said Privileged Access monitors identities and permissions across cloud, database and Kubernetes environments to identify access that should be removed or reduced. Requests and approvals are logged with attribution, which the company says can support audit evidence for SOC 2, HIPAA, PCI DSS, ISO 27001 and GDPR.
The workflow can approve lower-risk requests automatically based on policy. Requests deemed higher risk can be sent to reviewers through PagerDuty, Slack, Microsoft Teams or Jira, according to 1Password.
Separately, 1Password moved 1Password Credential Broker into public preview for GitHub Actions. The broker issues credentials tied to a single workflow run, so teams can avoid putting static secrets in pipeline configuration files. 1Password said the broker checks whether the requesting identity is trusted before delivering credentials and records the event.
Ben De St. Paer-Gotch, director of product management at GitHub Inc., said in the announcement that the integration lets engineering teams remove secrets from pipeline configurations while still giving GitHub Actions the access needed to build and ship software.
Developer security features reach general availability
1Password also made three developer-focused capabilities generally available in 1Password Enterprise Password Manager. Developer Watchtower detects exposed credentials in local .env files and prompts developers to move them into 1Password, while giving administrators visibility into credential risk on those machines.
1Password Environments allows developers to import existing .env files into the company’s vault and access secrets through 1Password’s Model Context Protocol server, which the company says keeps secrets off disk and out of model context. Credential Governance gives administrators a central view of company-owned credentials in employee and shared vaults, with controls to take ownership and manage access over time.
1Password said its enterprise vault stores more than 1.5 billion credentials and secrets. The company says it is used by more than 1 million developers and 180,000 businesses, including Canva, Cursor, Figma, GitHub, Hugging Face, MongoDB, Salesforce and Wiz.
This story draws on original reporting from SiliconANGLE.