Browser Extension Could Hijack Claude Desktop
Security researchers show a malicious extension running on claude.ai can make Claude Desktop execute code on a user's computer.
Updated on Nov. 6, 2026 in Artificial Intelligence
Live Poll
Do you trust AI tools with access to your computer to stay secure?
Forever Security researchers demonstrated that a browser extension able to run scripts on claude.ai could drive Claude Desktop to run code on a user's machine. The attack requires a person signed in to both claude.ai and Claude Desktop with Dispatch turned on.
Why it matters
Dispatch accepted an ordinary signed-in claude.ai session, letting an extension send Claude Desktop instructions as if the user had typed them. Anthropic accepted the report as valid but closed it as a duplicate of an issue it had already found internally.
In the test, a script opened Dispatch and sent four prompts as the user, after which a follow-up message got Claude to install an unfamiliar npm package. Approval prompts appeared in the web page the extension controlled, so the script could click them every second without a person watching.
The players
Forever Security
A security research firm that tested the Claude Desktop attack and reported the flaw to Anthropic.
Anthropic
The AI company that develops the Claude models and the Claude Desktop application, and that operates Dispatch.
The details
The researchers reported the flaw to Anthropic on May 14. In their test, Claude ran a global npm install, and the package's install script opened the Calculator app, showing that code execution was possible. Dispatch is now closed to new users, and its help page does not mention a retirement.
Timeline
May 14: Researchers reported the flaw to Anthropic.
May 19: Anthropic accepted the report as valid but closed it as a duplicate.
October 13: Anthropic told researchers Dispatch would be retired.
End of October: Target date for Dispatch retirement.
The Tech Race
The flaw centers on Dispatch, Anthropic's bridge between its web and desktop Claude products, which is slated for retirement with the end of October as the target. The episode underscores how AI assistants that act on a computer widen the security surface beyond the chat window itself.
Users who run Claude Desktop with Dispatch enabled while signed in to claude.ai could have instructions sent to their computer without their knowledge. Turning off Dispatch or avoiding unfamiliar browser extensions reduces exposure until the feature is retired.
The takeaway
Any AI assistant that can act on your computer inherits the trust of your signed-in sessions, so review which browser extensions and background integrations have access to them. Keep desktop AI features turned off when you do not need them.
What happens next
Dispatch is expected to be retired, with the end of October as the target date.
Further reading
For more on how AI tools are being built and secured, see the Artificial Intelligence section.
Source note: This article includes information reported by The Hacker News.
Live Poll
Do you trust AI tools with access to your computer to stay secure?
