# Anthropic details what stops Claude Duet from going rogue

DevFeed: [Anthropic details what stops Claude Duet from going rogue](<https://devfeed.tech/articles/anthropic-details-what-stops-claude-duet-from-going-rogue-100322.md>)

Original publisher: [Read original article](<https://www.helpnetsecurity.com/2026/10/26/anthropic-claude-duet-security-safeguards/>)

Author: Anamarija Pogorelec

Published: 2026-10-26T10:41:47Z

Content type: news

Language: en

Sources: [Help Net Security](<https://devfeed.tech/sources/help-net-security.md>)

Topics: [Claude](<https://devfeed.tech/topics/claude.md>), [ai security](<https://devfeed.tech/topics/ai-security.md>), [cloud security](<https://devfeed.tech/topics/cloud-security.md>), [Cryptography](<https://devfeed.tech/topics/cryptography.md>)

Tags: [1password](<https://devfeed.tech/tags/1password.md>), [account](<https://devfeed.tech/tags/account.md>), [actions](<https://devfeed.tech/tags/actions.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [apps](<https://devfeed.tech/tags/apps.md>), [browser](<https://devfeed.tech/tags/browser.md>), [business](<https://devfeed.tech/tags/business.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [claude-duet](<https://devfeed.tech/tags/claude-duet.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [credentials](<https://devfeed.tech/tags/credentials.md>), [cybersecurity](<https://devfeed.tech/tags/cybersecurity.md>), [news](<https://devfeed.tech/tags/news.md>), [privacy](<https://devfeed.tech/tags/privacy.md>), [proactive](<https://devfeed.tech/tags/proactive.md>), [security](<https://devfeed.tech/tags/security.md>), [security-and-privacy](<https://devfeed.tech/tags/security-and-privacy.md>), [tasks](<https://devfeed.tech/tags/tasks.md>), [users](<https://devfeed.tech/tags/users.md>)

## AI overview

Claude Duet requires approval for sensitive actions such as sending messages, payments, and deleting information, with a separate check comparing each proposed action to the user's intent. Some actions remain prohibited even with approval, and certain actions require approval every time. A classifier checks outgoing messages and browser uploads for sensitive data; Anthropic says credentials, full card and bank numbers, and government IDs must never be shared, while other private details require consent. The company acknowledges that these checks may miss sensitive data or flag harmless content. Duet's cloud browser runs in a dedicated virtual machine, and the agent cannot access passwords, cookies, or sign-in tokens; passwords entered in the Claude app are encrypted and sent directly to the browser. Anthropic says its evaluations include prompt-injection scenarios, red-team exercises, and an external penetration test, with further browser and sign-in-form testing underway. Memory can retain task context until reset or history deletion; disconnecting an app does not remove information already retained.

## Source excerpt

Anthropic has outlined the security and privacy protections behind Claude Duet, its proactive AI agent that can monitor connected accounts and complete tasks while users are away. The safeguards include approval requirements, checks for sensitive information and infrastructure that keeps credentials outside the model's reach. Duet works through a cloud-based computer and browser to prepare briefings, draft messages and handle business tasks. The standalone app is available through a research preview waitlist for Pro and ... More -> The post Anthropic details what stops Claude Duet from going rogue appeared first on Help Net Security.