Skip to content
NewsProduct launch

1Password opens password-free AI agent logins beyond Claude to Muse, ChatGPT, and Grok Bot

· by Pondero Newsdesk

The short version

1Password launched 1Password for Agents on October 22, 2026, extending the passwordless credential-sharing system it built for Claude in July to Meta's Muse, OpenAI's ChatGPT Work and Dots, and Grok Bot.

1Password opens password-free AI agent logins beyond Claude to Muse, ChatGPT, and Grok Bot

Banks are already flagging AI-agent logins as suspicious account activity, per Axios's reporting cited in TechCrunch's coverage of the launch, the same week 1Password widened the one tool built to make those logins traceable. On October 22, 2026, the password manager launched 1Password for Agents, extending the passwordless sign-in system it built for Claude in July to Meta's Muse, OpenAI's ChatGPT Work and Dots, Grok Bot, and Town's Townie assistant, per 1Password's press release.

What

When one of the supported agents reaches a sign-in page, it sends 1Password a digital key that verifies its identity rather than a request for the password itself, per 1Password's press release. The person using the agent sees which credential it wants and why, approves the request on their own device, and 1Password fills the password or one-time code directly into the sign-in form. The agent, its memory, and the partner company's own systems never receive the credential. 1Password said the Muse and Grok Bot integrations are "rolling out over the coming days across all platforms" and that the feature is included in every Individual and Family plan, per 1Password's product page. Anthropic's chief product officer, Ami Vora, said in the press release that "Claude Duet asks before it signs in anywhere, and soon 1Password will complete the sign-in with the passwords people have already set up, so they don't have to start over."

More agents sharing one vault raises the approval-fatigue stakes

The security model depends on a person actually reading each approval prompt instead of reflexively tapping yes, and that discipline gets harder to hold as the number of agents asking for logins grows from one, Claude, to five named partners plus nine more companies 1Password lists as "working on" integrations, per 1Password's press release. Anyone running several AI agents against the same accounts now has a single approval queue to watch rather than a scattered set of shared passwords, which is the upside 1Password is selling. A person who starts auto-approving agent logins out of habit undoes the protection the approval step was designed to provide.

Context and reactions

1Password's first version of this, 1Password for Claude, shipped in July 2026 for Claude in Chrome, approving each request with a biometric prompt and scoping access to a single browser session, per IT Brief Australia's report at launch. That release also introduced what 1Password called Agentic Mode, which restricts the browser extension to only the credentials approved for a given task while an agent controls the browser, and scans the page afterward to confirm nothing was left exposed. October's expansion reuses that same session-scoped, approval-gated design across the new agent partners rather than introducing a new mechanism, based on both companies' public descriptions of the two launches. OpenAI's head of app platform, Glen Coates, and SpaceXSI's head of product, Tido Carriero, each framed the integration in the press release as letting their agents sign into accounts customers already have saved in 1Password without the model ever seeing the credential.

What to watch next

1Password named nine more companies, including Perplexity, Ramp, and Browserbase, as partners "working on" further integrations without giving a shipping date, per the press release. Whether those ship, and whether the bank concerns Axios reported the same day push other password managers toward the same zero-credential-exposure design, are the open questions through the rest of 2026.

Sources