What changed
2026-07-29 14:43 UTC → 2026-07-29 21:09 UTC ·
added
removed
After acquiring Apono in mid‑June 2026, 1Password folded the just‑in‑time access‑management platform into Following its Unified Access suite. June 2026 acquisition of Apono, 1Password has been rolling out AI‑driven credential controls. In July the company launched it released Agentic Mode, letting allowing Anthropic’s Claude AI to request credentials passwords and TOTP codes without ever exposing passwords or MFA codes. A them; a macOS browser extension, 1Password for Claude, debuted on 14 July, enabling extension let Claude to complete logins while the vault stayed locked. A briefing on 18 July described the zero‑exposure architecture: remained locked, with each request requires gated by biometric user verification, the vault auto‑locks while Claude controls the browser, verification and access is revoked automatically when revoked after the task ends, the browser closes, ends or after a maximum of nine hours. On 21 July 1Password announced the company added an AI‑powered automatic‑login feature that works with Claude on macOS (Claude desktop app and Chrome). When Claude needs to sign in, it asks prompts the user for the specific credential; the user authorises the request with biometric verification, and authorisation before 1Password fills the login fields without ever revealing the password or TOTP code to the AI. The feature currently supports passwords and time‑based one‑time passwords, with plans to add payment cards and identity documents. On 27 July credentials for Claude on macOS. Two days later 1Password introduced Privileged Access, a solution just‑in‑time provisioning system that grants issues one‑time credentials to engineers and AI agents one‑time credentials for critical infrastructure cloud, database and automatically revokes them after a task completes. Built on Apono’s just‑in‑time provisioning, it writes permissions directly into native policy layers of cloud environments, databases Kubernetes resources and Kubernetes, ensuring credentials never leave the system. A company study found 40 % of developers give agents persistent access, a risk the new product aims to eliminate. Privileged Access records every request for compliance. On 29 July the solution was extended with privileged‑access limits for autonomous AI agents: permissions are scoped to individual tasks, low‑risk requests are auto‑approved while higher‑risk ones trigger human review via Slack, Teams or Jira, and approval, supporting compliance each AI receives its own identity with intent monitoring that can revoke access if actions diverge from the declared purpose. All actions remain logged for ISO 27001, PCI‑DSS, GDPR and U.S. health‑care regulations. compliance.