Password manager 1Password on Thursday launched a new capability that lets users permit AI agents to sign into their personal accounts on their behalf without the agents ever seeing the users' passwords or other credentials.
How 1Password for Agents works
Branded 1Password for Agents, the platform establishes a communication flow among the user, the AI agent, and the password manager. When an agent attempts to log in for a user, a unique digital key is sent to 1Password to verify the agent's digital identity. The user receives an alert about the login attempt and can personally approve the request. Once approved, 1Password signs the agent in by filling in the credentials on the assistant's behalf, enabling the agent to complete the requested task without gaining access to the underlying secrets.
The company says users can select their preferred agent, including Anthropic's Claude Duet, OpenAI's Dots and ChatGPT Work, Grok Bot, Meta's Muse, and agents from Town, an AI assistant startup. 1Password also stated it is working with other firms to expand support.
Why this matters
In a blog post, 1Password noted that handing over a password, secret, or credential is a common shortcut to enable an agent but can leave sensitive information in places users no longer control if copies are retained. 1Password for Agents aims to offer a trusted way to delegate tasks to personal AI agents while preserving user control over passwords, secrets, and credentials required for those tasks.
Context: banks are noticing agentic activity
The rollout coincides with an Axios report saying U.S. banks have begun detecting agentic activity inside customer accounts. That activity does not necessarily indicate rogue agents stealing account access; rather it reflects that consumers are increasingly allowing automated services to act on their behalf, including for financial matters.
Still, banks face challenges: many institutions cannot reliably determine whether agentic actions were initiated by users or occurred without their instruction. Bankers interviewed described the phenomenon as "horrifying," and a Chase spokesperson told Axios that customers "should not give their usernames and passwords to anyone."
Conclusion
1Password for Agents is positioned as a compromise between convenience and security, enabling users to grant AI assistants the ability to act on their behalf without exposing credentials. The launch arrives as financial institutions and security teams struggle to adapt to growing uses of automation and to distinguish legitimate user-authorized activity from potentially risky agent behavior.
