1Password launches 1Password for Agents, letting AI agents log in to user accounts without seeing credentials
1Password released "1Password for Agents" on Thursday. When an agent tries to log in, it sends a unique digital key to 1Password to confirm its identity. The user gets an alert and can approve the request. 1Password then fills in the credentials and signs the agent in, so the agent never sees the password. Supported agents include Anthropic's Claude Duet, OpenAI's Dots and ChatGPT Work, Grok Bot, Meta's Muse and agents from Town. The company says it is working with others to expand support.
Entities: 1Password, 1Password for Agents, Anthropic, Claude Duet, OpenAI, Dots
0 primary
What happened
1Password released "1Password for Agents" on Thursday. When an AI agent tries to log in to an account, it sends a unique digital key to 1Password to prove its identity. The user gets an alert and can approve the request, and 1Password then fills in the credentials and signs the agent in, so the agent never sees the password. Named supported agents include Anthropic's Claude Duet, OpenAI's Dots and ChatGPT Work, Grok Bot, Meta's Muse and agents from Town, with more promised. The coverage gives no pricing, availability dates or customer numbers.
Why it matters
Anyone letting agents act in their accounts currently has to either hand over passwords or avoid the task, so a approve-and-fill pattern is a sensible, practical step. Existing 1Password users and security teams at firms piloting agents are the most likely to benefit, and it gives them a concrete control to test. The impact is incremental: it is one vendor's integration, not a standard, and it only helps where agents and sites cooperate with it. The agent still gets a logged-in session once approved, so the risk is reduced rather than removed.
What is noise
The bank angle (US banks detecting agentic activity they cannot attribute) comes from Axios and is background, not something this launch demonstrates it solves. The claim that credentials are kept out of the agent's sight is the vendor's own description and has not been independently tested, and the report does not say what an agent can do after sign-in, how long sessions last, or how many agent makers have actually built support rather than merely being listed.
Watch next
- 01Pricing, plan tiers and general availability, plus whether this works for personal accounts or only business ones, within the next quarter.
- 02Independent security reviews or researcher write-ups testing whether credentials, session tokens or approval prompts can be leaked, spoofed or abused through the agent flow.
- 03Whether the agent makers confirm support themselves, and whether rivals such as Bitwarden, Dashlane or the browser and platform vendors ship equivalents or a shared standard.
Coverage
1 storyNewsletter
Get the no-hype AI news straight to your inbox
Weekly. No spam, unsubscribe any time. Privacy
More infrastructure signals
Full feed →- Deepseek releases V4.1-Flash, an open-source model that sharply cuts KV cache memory and input-processing compute for AI agents10 Sept 202682
- Meta releases Muse Spark Open, an open-weight ~816B-parameter multimodal mixture-of-experts model, with Hugging Face support15 Oct 202680
- OpenAI discloses sandbox-escape and credential-leak incidents, confirms pause on tool-use for its most capable models26 Sept 202680
- Anthropic threat report: Claude abused for malware, drone/missile software, mass surveillance, and industrial-scale distillation by Chinese AI labs11 Sept 202680